Daily Papers

Large Audio Language Models (LALMs) are prone to hallucinating and over-relying on text priors when simultaneously presented with audio and text inputs. To mitigate these hallucinations, we propose utilizing the multimodal Direct Preference Optimization (mDPO) objective, which forces the model to ground its generation in the acoustic input by contrasting intact and distorted audio counterparts. We extend this preference learning framework to the audio domain by applying a variety of acoustic perturbations. Evaluating the Qwen2-Audio backbone across the DCASE 2025 Challenge and AH Existence datasets, we demonstrate that extending mDPO to LALMs significantly enhances temporal reasoning in the complex DCASE dataset, and improves performance on basic existence verification in the AH benchmark. We identify temporal reversal, frequency masking, and random noise as the most effective perturbations. Ultimately, our approach achieves an absolute accuracy improvement of 14.0% on the DCASE 2025 dataset and 27.4% on AH Existence.

The goal of this paper is to provide a new perspective on audio-visual target speaker extraction (AV-TSE) by decoupling the separation and target selection. Conventional AV-TSE systems typically integrate audio and visual features deeply to re-learn the entire separation process, which can act as a fidelity ceiling due to the noisy nature of in-the-wild audio-visual datasets. To address this, we propose Plug-and-Steer, which assigns high-fidelity separation to a frozen audio-only backbone and limits the role of visual modality strictly to target selection. We introduce the Latent Steering Matrix (LSM), a minimalist linear transformation that re-routes latent features within the backbone to anchor the target speaker to a designated channel. Experiments across four representative architectures show that our method effectively preserves the acoustic priors of diverse backbones, achieving perceptual quality comparable to the original backbones. Audio samples are available at: https://plugandsteer.github.io

Text-to-audio-video (T2AV) generation models produce a video and its soundtrack from a textual description, but offer no control over whose voice speaks in the output. We show that a base T2AV model can be turned into a voice-cloning model by adding a single zero-initialized linear layer on top of its audio backbone, fine-tuning for a comparatively short training schedule, and conditioning on a short reference recording at inference time. The reference is injected through two complementary signals: its diffusion latents are prepended to the audio stream, and a global speaker embedding modulates token of the target audio. On a benchmark of 674 speaker-text pairs spanning 30 speakers we compare against five strong voice-cloning text-to-speech baselines: our enhanced 5B model attains the highest speaker-encoder cosine similarity (SECS) across three independent verification networks (ECAPA-TDNN, WavLM-SV, Resemblyzer), statistically significantly outperforming every baseline. A side product of the architecture is that the audio path can be evaluated without the video path at inference time, yielding a ~30x speed-up over the full audio-video diffusion loop while preserving the voice-cloning behaviour.

Audio source separation is fundamental for machines to understand complex acoustic environments and underpins numerous audio applications. Current supervised deep learning approaches, while powerful, are limited by the need for extensive, task-specific labeled data and struggle to generalize to the immense variability and open-set nature of real-world acoustic scenes. Inspired by the success of generative foundation models, we investigate whether pre-trained text-guided audio diffusion models can overcome these limitations. We make a surprising discovery: zero-shot source separation can be achieved purely through a pre-trained text-guided audio diffusion model under the right configuration. Our method, named ZeroSep, works by inverting the mixed audio into the diffusion model's latent space and then using text conditioning to guide the denoising process to recover individual sources. Without any task-specific training or fine-tuning, ZeroSep repurposes the generative diffusion model for a discriminative separation task and inherently supports open-set scenarios through its rich textual priors. ZeroSep is compatible with a variety of pre-trained text-guided audio diffusion backbones and delivers strong separation performance on multiple separation benchmarks, surpassing even supervised methods.

Video generation is progressing beyond isolated clips toward long-form narratives and interactive worlds, requiring models to preserve identities, follow user controls, and remain stable over extended rollouts. We present JoyAI-Echo-1.5, a unified audio-visual generation system with two purpose-built variants. The long-video variant introduces composable cross-shot memory that aggregates visual evidence across multiple prior shots and speaker cues derived from speech-filtered full-shot audio, enabling persistent character appearance and voice identity across flexible combinations of text, image, and memory conditioning. The world-model variant converts heterogeneous navigation inputs into calibrated metric 6-DoF camera trajectories and injects them through a geometry-aware conditioning pathway, enabling controller-agnostic interaction across flexible viewpoints. To support efficient long-horizon generation, we transform a bidirectional audio-visual backbone into a causal few-step generator using progressive teacher forcing and short- and long-horizon Self-Gradient Forcing on self-generated rollouts. Experiments demonstrate strong performance in both settings. JoyAI-Echo-1.5 achieves improvements over existing long-video baselines in cross-shot consistency, visual quality, text alignment, and speech fidelity. Its world-model variant ranks first on WBench, with an average score of 81.7, and achieves leading visual quality and long-horizon persistence on SANA-WM-Bench. Together, these results indicate that memory, geometric control, and rollout-aware training provide a practical foundation for generating coherent stories and continuously evolving interactive worlds. Project page: https://echo-team-joy-future-academy-jd.github.io/Echo-1.5-Page/.

Embodied assistants in domestic environments must infer what happened, where and when it occurred, and how to respond. To address this, we introduce RMS-AQA, a spatial audio question answering (SAQA) benchmark for real-world domestic environments. The benchmark features a two-stage question-answering (QA) format to comprehensively assess the ability of audio-language models (ALMs) to first ground audible sound events and subsequently perform complex spatio-temporal reasoning based on that grounding. To maximize acoustic realism, our dataset combines authentic real-world first-order Ambisonics (FOA) recordings with high-fidelity synthetic data generated using measured room impulse responses (RIRs). Furthermore, we provide a lightweight spatial plug-in that injects FOA-format data into frozen audio-language backbones. Experimental results reveal that the primary challenges stem from concurrent sources, far distance, and sim-to-real domain gap between RIR-synthesized and authentic recordings.

Omni-modal retrieval promises a single embedding space for text, image, video, document, and audio inputs, but building such a unified retriever is difficult since these modalities differ in data distribution, architecture, and optimization dynamics. In this work, we present Conan-embedding-v3, a decouple--fuse--recover framework for omni-modal retrieval. Conan-embedding-v3 first trains modality specialists independently and fuses their task vectors into a single dense backbone, a strategy we call Decoupled Specialist Fusion. We show that this fusion composes visual, video, and document retrieval capabilities, but also exposes a failure mode for projector-based modalities: when audio is attached through an external encoder and projector, fusing the backbone leaves the projector calibrated to the audio-specialist backbone, causing a large audio retrieval regression despite copying all audio-specific modules unchanged. We call this failure Projector Drift. To repair it, Conan-embedding-v3 applies Projector Recovery (i.e., full-parameter fine-tuning of the projector while keeping the backbone frozen) followed by balanced multi-modal rehearsal. The resulting model supports these retrieval pathways in one backbone, achieving 74.9 scores on MMEB while obtaining 55.61 on the 30-task MAEB audio suite.

Large language models (LLMs) have been widely used as knowledge backbones of Large Audio Language Models (LALMs), yet how much auditory knowledge they encode through text-only pre-training and how this affects downstream performance remains unclear. We study this gap by comparing different LLMs under two text-only and one audio-grounded setting: (1) direct probing on AKB-2000, a curated benchmark testing the breadth and depth of auditory knowledge; (2) cascade evaluation, where LLMs reason over text descriptions from an audio captioner; and (3) audio-grounded evaluation, where each LLM is fine-tuned into a Large Audio Language Model (LALM) with an audio encoder. Our findings reveal that auditory knowledge varies substantially across families, and text-only results are strongly correlated with audio performance. Our work provides empirical grounding for a comprehensive understanding of LLMs in audio research.

Advanced speech synthesis technologies have enabled highly realistic speech generation, posing security risks that motivate research into audio deepfake detection (ADD). While state space models (SSMs) offer linear complexity, pure causal SSMs architectures often struggle with the content-based retrieval required to capture global frequency-domain artifacts. To address this, we explore the scaling properties of hybrid architectures by proposing XLSR-MamBo, a modular framework integrating an XLSR front-end with synergistic Mamba-Attention backbones. We systematically evaluate four topological designs using advanced SSM variants, Mamba, Mamba2, Hydra, and Gated DeltaNet. Experimental results demonstrate that the MamBo-3-Hydra-N3 configuration achieves competitive performance compared to other state-of-the-art systems on the ASVspoof 2021 LA, DF, and In-the-Wild benchmarks. This performance benefits from Hydra's native bidirectional modeling, which captures holistic temporal dependencies more efficiently than the heuristic dual-branch strategies employed in prior works. Furthermore, evaluations on the DFADD dataset demonstrate robust generalization to unseen diffusion- and flow-matching-based synthesis methods. Crucially, our analysis reveals that scaling backbone depth effectively mitigates the performance variance and instability observed in shallower models. These results demonstrate the hybrid framework's ability to capture artifacts in spoofed speech signals, providing an effective method for ADD.

Historical films suffer from co-occurring visual and audio degradations---blur, noise, flicker, hiss, clipping, and dropout---yet existing methods restore each modality independently, leaving quality gaps and cross-modal inconsistency. We present OmniVR, the first joint audio-video generative restoration model. Built upon a 22B-parameter audio-video generation backbone, OmniVR formulates restoration as conditional generation within a unified multimodal DiT: the low-quality video and audio are encoded as latent conditions, combined with a fixed restoration prompt, and jointly denoised to recover visual structure, temporal motion, and acoustic detail under one coordinated objective. Three key designs enable this adaptation: (1) a joint audio-video degradation pipeline that simulates real old-film characteristics from Internet-collected data; (2) an architecture-preserving text-to-audio-video (T2AV) to audio-video-to-audio-video (AV2AV) transition with prompt annealing that maximally retains the generative prior; and (3) first-frame image-to-video (I2V) anchoring with loss reweighting and waveform supervision for long-video extrapolation and audio fidelity. We also propose OmniVRBench, the first benchmark that evaluates audio-video restoration across visual quality, audio quality, temporal consistency, and audio-visual synchrony on 200 real historical clips. OmniVR surpasses all prior methods on all six visual metrics, achieves the best audio quality, and produces natural colorization---the first method to jointly address all three aspects. Code and weights will be publicly released. Project Page: https://xin1u.github.io/OminiVR_PAGE/

Traditional audio-visual methods rely on independent audio and visual backbones, which is costly and not scalable. In this work, we investigate using an audio-visual siamese network (AVSiam) for efficient and scalable audio-visual pretraining. Our framework uses a single shared vision transformer backbone to process audio and visual inputs, improving its parameter efficiency, reducing the GPU memory footprint, and allowing us to scale our method to larger datasets and model sizes. We pretrain our model using a contrastive audio-visual matching objective with a multi-ratio random masking scheme, which enables our model to process larger audio-visual instance batches, helpful for contrastive learning. Unlike prior audio-visual methods, our method can robustly handle audio, visual, and audio-visual inputs with a single shared ViT backbone. Furthermore, despite using the shared backbone for both modalities, AVSiam achieves competitive or even better results than prior methods on AudioSet and VGGSound for audio-visual classification and retrieval. Our code is available at https://github.com/GenjiB/AVSiam

We introduce Music Flamingo, a novel large audio-language model designed to advance music (including song) understanding in foundational audio models. While audio-language research has progressed rapidly, music remains challenging due to its dynamic, layered, and information-dense nature. Progress has been further limited by the difficulty of scaling open audio understanding models, primarily because of the scarcity of high-quality music data and annotations. As a result, prior models are restricted to producing short, high-level captions, answering only surface-level questions, and showing limited generalization across diverse musical cultures. To address these challenges, we curate MF-Skills, a large-scale dataset labeled through a multi-stage pipeline that yields rich captions and question-answer pairs covering harmony, structure, timbre, lyrics, and cultural context. We fine-tune an enhanced Audio Flamingo 3 backbone on MF-Skills and further strengthen multiple skills relevant to music understanding. To improve the model's reasoning abilities, we introduce a post-training recipe: we first cold-start with MF-Think, a novel chain-of-thought dataset grounded in music theory, followed by GRPO-based reinforcement learning with custom rewards. Music Flamingo achieves state-of-the-art results across 10+ benchmarks for music understanding and reasoning, establishing itself as a generalist and musically intelligent audio-language model. Beyond strong empirical results, Music Flamingo sets a new standard for advanced music understanding by demonstrating how models can move from surface-level recognition toward layered, human-like perception of songs. We believe this work provides both a benchmark and a foundation for the community to build the next generation of models that engage with music as meaningfully as humans do.

Existing video personalization methods preserve visual likeness but treat video and audio separately. Without access to the visual scene, audio models cannot synchronize sounds with on-screen actions; and because classical voice-cloning models condition only on a reference recording, a text prompt cannot redirect speaking style or acoustic environment. We propose ID-LoRA (Identity-Driven In-Context LoRA), which jointly generates a subject's appearance and voice in a single model, letting a text prompt, a reference image, and a short audio clip govern both modalities together. ID-LoRA adapts the LTX-2 joint audio-video diffusion backbone via parameter-efficient In-Context LoRA and, to our knowledge, is the first method to personalize visual appearance and voice in a single generative pass. Two challenges arise. Reference and generation tokens share the same positional-encoding space, making them hard to distinguish; we address this with negative temporal positions, placing reference tokens in a disjoint RoPE region while preserving their internal temporal structure. Speaker characteristics also tend to be diluted during denoising; we introduce identity guidance, a classifier-free guidance variant that amplifies speaker-specific features by contrasting predictions with and without the reference signal. In human preference studies, ID-LoRA is preferred over Kling 2.6 Pro by 73% of annotators for voice similarity and 65% for speaking style. On cross-environment settings, speaker similarity improves by 24% over Kling, with the gap widening as conditions diverge. A preliminary user study further suggests that joint generation provides a useful inductive bias for physically grounded sound synthesis. ID-LoRA achieves these results with only ~3K training pairs on a single GPU. Code, models, and data will be released.

Audio-text retrieval is crucial for bridging acoustic signals and natural language. While contrastive dual-encoder architectures like CLAP have shown promise, they are fundamentally limited by the capacity of small-scale encoders. Specifically, the text encoders struggle to understand complex queries that require reasoning or world knowledge. In this paper, we propose AuroLA, a novel contrastive language-audio pre-training framework that re-purposes Multimodal Large Language Models (MLLMs) as a unified backbone for retrieval. Specifically, we make three contributions: (i) we construct a scalable data pipeline that curates diverse audio from multiple sources and generates multi-granular captions, ranging from long descriptions to structured tags, via automated annotation; (ii) we adapt an MLLM for retrieval by prompting it to summarize the audio/text input and using the hidden state of a special token as audio/text embeddings. For model training, we devise a novel Hybrid-NCE loss, which employs multi-granular supervision and hard-negative reweighting to robustly align audio with diverse textual supervision; and (iii) we design an MLLM-based bidirectional re-ranking module that refines retrieval candidates through deep cross-modal interaction. Extensive experiments demonstrate that AuroLA consistently outperforms state-of-the-art models, including the recent PE-AV, while utilizing only approximately 1% of PE-AV's training data. Lastly, we observe clear scaling trends regarding dataset size and model capacity, validating the effectiveness of MLLM as a unified backbone for audio-text retrieval. Code is available at https://github.com/Jazzcharles/AuroLA.

DCASE~2026 Task~5 introduces Audio-Dependent Question Answering (ADQA), which tests whether large audio-language models answer from the audio rather than from textual priors. An Audio-Dependency Filtering (ADF) pipeline combines silent-audio probing, per-option perplexity, a large language model (LLM) commonsense check, and human review to remove items solvable from text alone. The 3000 items that pass form the ADQA-Bench evaluation set, spanning music, speech, and environmental audio. The inaugural edition draws 14 teams and 36 submissions across two tracks defined by total parameter count (up to 100B and under 10B). A Chung-Ang University ensemble of MOSS-Audio-8B-Thinking and Qwen3-Omni-30B reaches the top overall accuracy at 58.33, and a MOSS-only configuration from the same team leads the sub-10B track at 57.30. Across the 30 submissions with a comparable development score, evaluation accuracy falls by 11.91 percentage points (pp) on average (median 10.91\,pp) on the hidden evaluation split, which is designed to be harder than the development split. The most common building blocks are: the MOSS-Audio-8B-Thinking backbone (13 of 36 submissions), Low-Rank Adaptation (LoRA) fine-tuning on AudioMCQ-StrongAC, and preference or reinforcement-learning objectives -- Group Relative Policy Optimization (GRPO) in five teams, Group reward-Decoupled Normalization Policy Optimization (GDPO) in two. At test time, prompt engineering is near-universal, and majority or choice-permutation voting is common. Every system misses the same set of 233 evaluation items.

Recent diffusion-based methods have achieved impressive progress in video content manipulation. However, they typically ignore the accompanying audio, leaving the audio disjointed from the edited results. In this paper, we propose InstructAV2AV, the first end-to-end framework for instruction-guided audio-video joint editing. We first develop a scalable data synthesis pipeline and construct InsAVE-80K, the first large-scale audio-video editing dataset with high-quality source-to-target pairs. With this data foundation, we adapt an audio-video generation backbone to leverage its robust priors. We concatenate the audio-video input with noisy latent codes to anchor the source context, propose the source-instruction gated attention to improve instruction following and content preservation, and introduce a two-stage training strategy to effectively transfer these pre-trained priors. Extensive experiments demonstrate that InstructAV2AV outperforms state-of-the-art methods across 11 metrics spanning three aspects on two evaluation sets, highlighting its potential for controllable content creation. Project page: https://hjzheng.net/projects/InstructAV2AV/.

Audio-video generation has often relied on complex multi-stage architectures or sequential synthesis of sound and visuals. We introduce Ovi, a unified paradigm for audio-video generation that models the two modalities as a single generative process. By using blockwise cross-modal fusion of twin-DiT modules, Ovi achieves natural synchronization and removes the need for separate pipelines or post hoc alignment. To facilitate fine-grained multimodal fusion modeling, we initialize an audio tower with an architecture identical to that of a strong pretrained video model. Trained from scratch on hundreds of thousands of hours of raw audio, the audio tower learns to generate realistic sound effects, as well as speech that conveys rich speaker identity and emotion. Fusion is obtained by jointly training the identical video and audio towers via blockwise exchange of timing (via scaled-RoPE embeddings) and semantics (through bidirectional cross-attention) on a vast video corpus. Our model enables cinematic storytelling with natural speech and accurate, context-matched sound effects, producing movie-grade video clips. All the demos, code and model weights are published at https://aaxwaz.github.io/Ovi

Self-supervised learning (SSL) underpins modern audio deepfake detection, yet most prior work centers on a single large wav2vec2-XLSR backbone, leaving compact under studied. We present RAPTOR, Representation Aware Pairwise-gated Transformer for Out-of-domain Recognition a controlled study of compact SSL backbones from the HuBERT and WavLM within a unified pairwise-gated fusion detector, evaluated across 14 cross-domain benchmarks. We show that multilingual HuBERT pre-training is the primary driver of cross-domain robustness, enabling 100M models to match larger and commercial systems. Beyond EER, we introduce a test-time augmentation protocol with perturbation-based aleatoric uncertainty to expose calibration differences invisible to standard metrics: WavLM variants exhibit overconfident miscalibration under perturbation, whereas iterative mHuBERT remains stable. These findings indicate that SSL pre-training trajectory, not model scale, drives reliable audio deepfake detection.

We introduce LMCodec, a causal neural speech codec that provides high quality audio at very low bitrates. The backbone of the system is a causal convolutional codec that encodes audio into a hierarchy of coarse-to-fine tokens using residual vector quantization. LMCodec trains a Transformer language model to predict the fine tokens from the coarse ones in a generative fashion, allowing for the transmission of fewer codes. A second Transformer predicts the uncertainty of the next codes given the past transmitted codes, and is used to perform conditional entropy coding. A MUSHRA subjective test was conducted and shows that the quality is comparable to reference codecs at higher bitrates. Example audio is available at https://mjenrungrot.github.io/chrome-media-audio-papers/publications/lmcodec.

We introduce StepAudio 3 Gen, a general-purpose audio generation model that supports zero-shot text-to-speech (TTS), voice design, vocal generation, sound effects, music, vibe speech, and mixtures of multiple audio types within a unified framework. At its core, StepAudio 3 Gen is a discrete autoregressive generator that models audio directly over residual vector quantization (RVQ) tokens, departing from the diffusion Transformer-based continuous generation paradigm prevalent in recent general audio models. Its StepAudio Tokenizer represents general audio at 12.5 Hz in a shared 16 times 2048 residual code space, jointly quantizing semantic and waveform-level acoustic features so that each code layer preserves both types of information. For generation, the backbone predicts the first codebook along the time axis using autoregressive modeling, while a lightweight causal Transformer completes the remaining fifteen codebooks along the codebook axis. Our study further identifies three key design principles: (1) interference-aware progressive pretraining for acquiring audio capabilities while preserving the textual abilities of the large language model, (2) RVQ Adaptor for effectively incorporating multi-codebook acoustic representations, and (3) discrete autoregressive modeling over a shared representation across general audio domains. With progressive pretraining, multi-task instruction training, and supervised fine-tuning, StepAudio 3 Gen achieves state-of-the-art performance on both TTS and voice design, while retaining strong generation capabilities across speech, vocals, sound effects, and music. Audio samples are available at https://stepaudiollm.github.io/step-audio-3-gen/.

We introduce Fish Audio Drama 3, an open-source speech generation model that turns text into directed performances, built on a 35B-parameter mixture-of-experts backbone that activates 3B parameters per token. It clones voices from multiple references, follows natural-language direction, revises a take when the next turn gives a note, voices multi-character scenes, edits existing speech by infilling, and synthesizes long-form and streaming audio. Multi-turn revision, scene dubbing, editing by infilling, and streaming text input are new in Drama 3, all in one model. Drama 3 pairs the hybrid linear-attention backbone with a new codebook multi-token-prediction audio decoder that predicts the acoustic codebooks of each frame in dedicated layers that attend over the entire sequence. The model is trained with a three-stage pre-training recipe, instruction fine-tuning, and large-batch reinforcement learning whose only reward is a single learned model of human preference. Its inference engine decodes a single stream at 271 frames per second on one H200, against 177 for Fish Audio S2, although Drama 3 has more than seven times as many parameters. The model weights are available on GitHub (https://github.com/fishaudio/fish-speech) and Hugging Face (https://huggingface.co/fishaudio/drama-3), and inference is available through SGLang-Omni (https://github.com/sgl-project/sglang-omni). We highly encourage readers to visit https://fish.audio to direct their own performances.

Prevailing Video-to-Audio (V2A) generation models operate offline, assuming an entire video sequence or chunks of frames are available beforehand. This critically limits their use in interactive applications such as live content creation and emerging generative world models. To address this gap, we introduce the novel task of frame-level online V2A generation, where a model autoregressively generates audio from video without access to future video frames. Furthermore, we propose SoundReactor, which, to the best of our knowledge, is the first simple yet effective framework explicitly tailored for this task. Our design enforces end-to-end causality and targets low per-frame latency with audio-visual synchronization. Our model's backbone is a decoder-only causal transformer over continuous audio latents. For vision conditioning, it leverages grid (patch) features extracted from the smallest variant of the DINOv2 vision encoder, which are aggregated into a single token per frame to maintain end-to-end causality and efficiency. The model is trained through a diffusion pre-training followed by consistency fine-tuning to accelerate the diffusion head decoding. On a benchmark of diverse gameplay videos from AAA titles, our model successfully generates semantically and temporally aligned, high-quality full-band stereo audio, validated by both objective and human evaluations. Furthermore, our model achieves low per-frame waveform-level latency (26.3ms with the head NFE=1, 31.5ms with NFE=4) on 30FPS, 480p videos using a single H100. Demo samples are available at https://koichi-saito-sony.github.io/soundreactor/.

Recent Diffusion Transformers (DiTs) have shown impressive capabilities in generating high-quality single-modality content, including images, videos, and audio. However, it is still under-explored whether the transformer-based diffuser can efficiently denoise the Gaussian noises towards superb multimodal content creation. To bridge this gap, we introduce AV-DiT, a novel and efficient audio-visual diffusion transformer designed to generate high-quality, realistic videos with both visual and audio tracks. To minimize model complexity and computational costs, AV-DiT utilizes a shared DiT backbone pre-trained on image-only data, with only lightweight, newly inserted adapters being trainable. This shared backbone facilitates both audio and video generation. Specifically, the video branch incorporates a trainable temporal attention layer into a frozen pre-trained DiT block for temporal consistency. Additionally, a small number of trainable parameters adapt the image-based DiT block for audio generation. An extra shared DiT block, equipped with lightweight parameters, facilitates feature interaction between audio and visual modalities, ensuring alignment. Extensive experiments on the AIST++ and Landscape datasets demonstrate that AV-DiT achieves state-of-the-art performance in joint audio-visual generation with significantly fewer tunable parameters. Furthermore, our results highlight that a single shared image generative backbone with modality-specific adaptations is sufficient for constructing a joint audio-video generator. Our source code and pre-trained models will be released.

Building on the Joint-Embedding Predictive Architecture (JEPA) paradigm, a recent self-supervised learning framework that predicts latent representations of masked regions in high-level feature spaces, we propose Audio-JEPA (Audio Joint-Embedding Predictive Architecture), tailored specifically for audio data. Audio-JEPA uses a simple Vision Transformer backbone to predict latent representations of masked spectrogram patches rather than reconstructing raw audio. We pre-train on unlabeled AudioSet clips (10s, 32kHz) with random patch masking on mel-spectrograms. We evaluate on the X-ARES suite covering speech, music, and environmental sound tasks. Although our implementation is a straightforward translation of the original model to audio, the results still show comparable performance to wav2vec 2.0 and data2vec while using less than one-fifth of their training data and with no hyper-parameter tuning. All code and pretrained checkpoints will be released on GitHub.

Audio Language Models (ALM) have emerged as the dominant paradigm for speech and music generation by representing audio as sequences of discrete tokens. Yet, unlike text tokens, which are invertible, audio tokens are extracted from lossy codecs with a limited bitrate. As a consequence, increasing audio quality requires generating more tokens, which imposes a trade-off between fidelity and computational cost. We address this issue by studying Continuous Audio Language Models (CALM). These models instantiate a large Transformer backbone that produces a contextual embedding at every timestep. This sequential information then conditions an MLP that generates the next continuous frame of an audio VAE through consistency modeling. By avoiding lossy compression, CALM achieves higher quality at lower computational cost than their discrete counterpart. Experiments on speech and music demonstrate improved efficiency and fidelity over state-of-the-art discrete audio language models, facilitating lightweight, high-quality audio generation. Samples are available at https://continuous-audio-language-models.github.io

Current audio language models are predominantly text-first, either extending pre-trained text LLM backbones or relying on semantic-only audio tokens, limiting general audio modeling. This paper presents a systematic empirical study of native audio foundation models that apply next-token prediction to audio at scale, jointly modeling semantic content, acoustic details, and text to support both general audio generation and cross-modal capabilities. We provide comprehensive empirical insights for building such models: (1) We systematically investigate design choices -- data sources, text mixture ratios, and token composition -- establishing a validated training recipe. (2) We conduct the first scaling law study for discrete audio models via IsoFLOP analysis on 64 models spanning 3{times}10^{18} to 3{times}10^{20} FLOPs, finding that optimal data grows 1.6times faster than optimal model size. (3) We apply these lessons to train SODA (Scaling Open Discrete Audio), a suite of models from 135M to 4B parameters on 500B tokens, comparing against our scaling predictions and existing models. SODA serves as a flexible backbone for diverse audio/text tasks -- we demonstrate this by fine-tuning for voice-preserving speech-to-speech translation, using the same unified architecture.

Talking-video character replacement requires coordinated transfer of appearance and voice while preserving the source motion, scene, linguistic content, and audio-video timing. Existing methods use separately optimized models for the two modalities, making audio-visual consistency difficult to enforce. We present UniSwap, the first framework for streaming joint audio-visual identity replacement in talking videos. Given a source video, a reference image, and a reference voice clip, UniSwap transfers the reference appearance and vocal timbre within a single audio-visual diffusion transformer while preserving the source content and dynamics. To address the scarcity of aligned cross-identity training pairs, we introduce a swap-and-reconstruct pipeline that removes visual and vocal identity from real clips and uses the original clips as reconstruction targets. Starting from a bidirectional backbone, we progressively adapt the model through In-context Pretraining for joint replacement, Conditional Streaming Adaptation for block-causal KV-cached generation, and Efficient Self-forcing DMD for mitigating exposure bias and reducing sampling from 30 to 3 denoising steps per block. Efficient Multi-LoRA Switching enables the three DMD roles to share a single frozen backbone. Feature-RoPE Decomposition keeps cached positions within the training range, supporting stable long-form inference. Experiments demonstrate strong audio-visual synchronization, competitive identity preservation, efficient streaming, and stable long-form generation.

Reducing audio-encoder depth lowers the inference cost of speech large language models, but removing complete blocks perturbs the embeddings consumed by the decoder and can cause deletion and premature end-of-sequence errors. We introduce X-AuT, a progressive framework that selects layer combinations through short behavioral probes and restores the pruned model through representation alignment, cross-scale distillation, scheduled student-policy supervision, and LoRA finetuning. The language-model backbone remains frozen, while attention LoRA adapters and the tied output embedding adapt during distillation. Training uses the highest-agreement tier from a transcript-consistency pipeline, followed by source reweighting during finetuning. On ten public Chinese--English benchmarks, compressing Qwen3-ASR-0.6B from 18 to 16 audio-encoder layers reduces macro-average error from 5.61% to 5.27%. The 14-layer model reaches 5.75% with 20.7% fewer audio-tower parameters. Under the matched recipe, the 1.7B teacher yields 5.55% mean error, compared with 8.45% for self-distillation, and progressive 18rightarrow14 pruning outperforms direct pruning (5.75% vs. 6.73%). These single-run results establish two practical operating points and show that the accuracy effects vary across benchmarks. Project website: https://xpeng-ai.github.io/x-aut

In recent years, automatic speech recognition (ASR) has witnessed transformative advancements driven by three complementary paradigms: data scaling, model scaling, and deep integration with large language models (LLMs). However, bridging the gap between academic benchmark performance and real-world production utility remains a persistent challenge, particularly in handling diverse regional dialects, dynamic entities and hotwords, long-range contextual information, and disfluent spontaneous speech. In this report, we present Qwen-Audio-3.0-ASR, a Mixture-of-Experts (MoE) LLM-based ASR system designed to address these production demands through a unified, instruction-following framework. The model is built upon the Qwen backbone, and is trained on tens of millions of hours of large-scale speech data. Qwen-Audio-3.0-ASR supports transcription across 30 languages and 16 Chinese dialectal varieties spanning eight major dialect regions. Beyond multilingual and dialectal recognition, the model provides production-oriented capabilities including industry-domain entity recognition, hierarchical hotword customization, native single-pass transcription polishing, and long-audio contextual modeling. We further develop a dedicated streaming variant, Qwen-Audio-3.0-ASR-Streaming, for latency-sensitive applications. Extensive evaluations on Chinese, English, multilingual, and real-world industrial test sets demonstrate state-of-the-art or highly competitive recognition performance across a broad range of evaluation conditions, with strong performance relative to leading commercial and proprietary systems including GPT-4o Transcribe and Gemini 3.1 Pro.

Audio deepfakes are a growing challenge for the general public, as well as for journalists and fact-checkers. The latter need reliable tools to verify the authenticity of their sources, while at the same time keeping their information private. Commercial deepfake detection solutions rely on cloud-based processing, which raises privacy concerns. To solve this problem, we propose an on-device audio deepfake detection model. We show that a truncated self-supervised backbone with a simple logistic classifier is both very fast and often more accurate than existing solutions. Our solution outperforms the baseline AASIST by 10% and improves inference speed by 40%. We integrate this model into a browser plug-in, which allows journalists and fact-checkers to detect deepfakes easily and securely. Code for the plugin is available at https://github.com/OctavianPascu97/Audio-Deepfakes-Browser-Plugin.

Current diffusion models for audio-driven avatar video generation struggle to synthesize long videos with natural audio synchronization and identity consistency. This paper presents StableAvatar, the first end-to-end video diffusion transformer that synthesizes infinite-length high-quality videos without post-processing. Conditioned on a reference image and audio, StableAvatar integrates tailored training and inference modules to enable infinite-length video generation. We observe that the main reason preventing existing models from generating long videos lies in their audio modeling. They typically rely on third-party off-the-shelf extractors to obtain audio embeddings, which are then directly injected into the diffusion model via cross-attention. Since current diffusion backbones lack any audio-related priors, this approach causes severe latent distribution error accumulation across video clips, leading the latent distribution of subsequent segments to drift away from the optimal distribution gradually. To address this, StableAvatar introduces a novel Time-step-aware Audio Adapter that prevents error accumulation via time-step-aware modulation. During inference, we propose a novel Audio Native Guidance Mechanism to further enhance the audio synchronization by leveraging the diffusion's own evolving joint audio-latent prediction as a dynamic guidance signal. To enhance the smoothness of the infinite-length videos, we introduce a Dynamic Weighted Sliding-window Strategy that fuses latent over time. Experiments on benchmarks show the effectiveness of StableAvatar both qualitatively and quantitatively.

Text-to-video and image-to-video generation have made rapid progress in visual quality, but they remain limited in controlling the precise timing of motion. In contrast, audio provides temporal cues aligned with video motion, making it a promising condition for temporally controlled video generation. However, existing audio-to-video (A2V) models struggle with fine-grained synchronization due to indirect conditioning mechanisms or limited temporal modeling capacity. We present Syncphony, which generates 380x640 resolution, 24fps videos synchronized with diverse audio inputs. Our approach builds upon a pre-trained video backbone and incorporates two key components to improve synchronization: (1) Motion-aware Loss, which emphasizes learning at high-motion regions; (2) Audio Sync Guidance, which guides the full model using a visually aligned off-sync model without audio layers to better exploit audio cues at inference while maintaining visual quality. To evaluate synchronization, we propose CycleSync, a video-to-audio-based metric that measures the amount of motion cues in the generated video to reconstruct the original audio. Experiments on AVSync15 and The Greatest Hits datasets demonstrate that Syncphony outperforms existing methods in both synchronization accuracy and visual quality. Project page is available at: https://jibin86.github.io/syncphony_project_page

Training a unified model integrating video-to-audio (V2A), text-to-audio (T2A), and joint video-text-to-audio (VT2A) generation offers significant application flexibility, yet faces two unexplored foundational challenges: (1) the scarcity of high-quality audio captions with tight A-V-T alignment, leading to severe semantic conflict between multimodal conditions, and (2) cross-task and intra-task competition, manifesting as an adverse V2A-T2A performance trade-off and modality bias in the VT2A task. First, to address data scarcity, we introduce SoundAtlas, a large-scale dataset (470k pairs) that significantly outperforms existing benchmarks and even human experts in quality. Powered by a novel agentic pipeline, it integrates Vision-to-Language Compression to mitigate visual bias of MLLMs, a Junior-Senior Agent Handoff for a 5 times cost reduction, and rigorous Post-hoc Filtering to ensure fidelity. Consequently, SoundAtlas delivers semantically rich and temporally detailed captions with tight V-A-T alignment. Second, we propose Omni2Sound, a unified VT2A diffusion model supporting flexible input modalities. To resolve the inherent cross-task and intra-task competition, we design a three-stage multi-task progressive training schedule that converts cross-task competition into joint optimization and mitigates modality bias in the VT2A task, maintaining both audio-visual alignment and off-screen audio generation faithfulness. Finally, we construct VGGSound-Omni, a comprehensive benchmark for unified evaluation, including challenging off-screen tracks. With a standard DiT backbone, Omni2Sound achieves unified SOTA performance across all three tasks within a single model, demonstrating strong generalization across benchmarks with heterogeneous input conditions. The project page is at https://swapforward.github.io/Omni2Sound.

Listening to heart and lung sounds - auscultation - is one of the first and most fundamental steps in a clinical examination. Despite being fast and non-invasive, it demands years of experience to interpret subtle audio cues. Recent deep learning methods have made progress in automating cardiopulmonary sound analysis, yet most are restricted to simple classification and offer little clinical interpretability or decision support. We present StethoLM, the first audio-language model specialized for cardiopulmonary auscultation, capable of performing instruction-driven clinical tasks across the full spectrum of auscultation analysis. StethoLM integrates audio encoding with a medical language model backbone and is trained on StethoBench, a comprehensive benchmark comprising 77,027 instruction-response pairs synthesized from 16,125 labeled cardiopulmonary recordings spanning seven clinical task categories: binary classification, detection, reporting, reasoning, differential diagnosis, comparison, and location-based analysis. Through multi-stage training that combines supervised fine-tuning and direct preference optimization, StethoLM achieves substantial gains in performance and robustness on out-of-distribution data. Our work establishes a foundation for instruction-following AI systems in clinical auscultation.

Large Audio-Language Models (LALMs) are becoming essential as a powerful multimodal backbone for real-world applications. However, recent studies show that audio inputs can more easily elicit harmful responses than text, exposing new risks toward deployment. While safety alignment has made initial advances in LLMs and Large Vision-Language Models (LVLMs), we find that vanilla adaptation of these approaches to LALMs faces two key limitations: 1) LLM-based steering fails under audio input due to the large distributional gap between activations, and 2) prompt-based defenses induce over-refusals on benign-speech queries. To address these challenges, we propose Safe-Ablated Refusal Steering (SARSteer), the first inference-time defense framework for LALMs. Specifically, SARSteer leverages text-derived refusal steering to enforce rejection without manipulating audio inputs and introduces decomposed safe-space ablation to mitigate over-refusal. Extensive experiments demonstrate that SARSteer significantly improves harmful-query refusal while preserving benign responses, establishing a principled step toward safety alignment in LALMs.

Spatial audio authoring coordinates source decisions with soundfield synthesis. For editing, plans specify what to change and retain, while reference audio supplies further acoustic detail. We present AMBIT, a unified framework for native First-Order Ambisonics (FOA) generation and editing with three connected contributions. First, a shared Transfusion backbone couples autoregressive planning with rectified-flow rendering through executable ScenePlans that specify source content, activity, and 3D trajectories. For editing, the plan describes the full target scene, including retained sources; contrastive features ground the requested change, while full reference latents supply residual acoustic detail for preservation. Second, an FOA-aware continuous VAE combines shared-decoder omnidirectional recovery, asymmetric grouped KL, and spatial covariance matching to preserve content and inter-channel structure. Third, joint Transfusion on-policy self-distillation (OPSD) refines planning on student-generated prefixes and rendering at sampling states reached under student plans, with target-conditioned teachers supervising both modes of the shared student. Evaluations across speech, music, and sound show improved generation content and spatial quality, acoustic fidelity during instruction-guided editing, and unedited-source preservation over the compared baselines. Project page: https://wjm-george.github.io/AMBIT/

We introduce Diffusion-based Audio Captioning (DAC), a non-autoregressive diffusion model tailored for diverse and efficient audio captioning. Although existing captioning models relying on language backbones have achieved remarkable success in various captioning tasks, their insufficient performance in terms of generation speed and diversity impede progress in audio understanding and multimedia applications. Our diffusion-based framework offers unique advantages stemming from its inherent stochasticity and holistic context modeling in captioning. Through rigorous evaluation, we demonstrate that DAC not only achieves SOTA performance levels compared to existing benchmarks in the caption quality, but also significantly outperforms them in terms of generation speed and diversity. The success of DAC illustrates that text generation can also be seamlessly integrated with audio and visual generation tasks using a diffusion backbone, paving the way for a unified, audio-related generative model across different modalities.

Foley Control is a lightweight approach to video-guided Foley that keeps pretrained single-modality models frozen and learns only a small cross-attention bridge between them. We connect V-JEPA2 video embeddings to a frozen Stable Audio Open DiT text-to-audio (T2A) model by inserting compact video cross-attention after the model's existing text cross-attention, so prompts set global semantics while video refines timing and local dynamics. The frozen backbones retain strong marginals (video; audio given text) and the bridge learns the audio-video dependency needed for synchronization -- without retraining the audio prior. To cut memory and stabilize training, we pool video tokens before conditioning. On curated video-audio benchmarks, Foley Control delivers competitive temporal and semantic alignment with far fewer trainable parameters than recent multi-modal systems, while preserving prompt-driven controllability and production-friendly modularity (swap/upgrade encoders or the T2A backbone without end-to-end retraining). Although we focus on Video-to-Foley, the same bridge design can potentially extend to other audio modalities (e.g., speech).

Generating realistic talking-head videos remains challenging due to persistent issues such as imperfect lip synchronization, unnatural motion, and evaluation metrics that correlate poorly with human perception. We propose FlowPortrait, a reinforcement-learning framework for audio-driven portrait animation built on a multimodal backbone for autoregressive audio-to-video generation. FlowPortrait introduces a human-aligned evaluation system based on Multimodal Large Language Models (MLLMs) to assess lip-sync accuracy, expressiveness, and motion quality. These signals are combined with perceptual and temporal consistency regularizers to form a stable composite reward, which is used to post-train the generator via Group Relative Policy Optimization (GRPO). Extensive experiments, including both automatic evaluations and human preference studies, demonstrate that FlowPortrait consistently produces higher-quality talking-head videos, highlighting the effectiveness of reinforcement learning for portrait animation.

We present AIBA (Attention-In-Band Alignment), a lightweight, training-free pipeline to quantify where text-to-audio diffusion models attend on the time-frequency (T-F) plane. AIBA (i) hooks cross-attention at inference to record attention probabilities without modifying weights; (ii) projects them to fixed-size mel grids that are directly comparable to audio energy; and (iii) scores agreement with instrument-band ground truth via interpretable metrics (T-F IoU/AP, frequency-profile correlation, and a pointing game). On Slakh2100 with an AudioLDM2 backbone, AIBA reveals consistent instrument-dependent trends (e.g., bass favoring low bands) and achieves high precision with moderate recall.

We present Eureka-Audio, a compact yet high-performance audio language model that achieves competitive performance against models that are 4 to 18 times larger across a broad range of audio understanding benchmarks. Despite containing only 1.7B parameters, Eureka-Audio demonstrates strong performance on automatic speech recognition (ASR), audio understanding, and dense audio captioning, matching or surpassing multiple 7B to 30B audio and omni-modal baselines. The model adopts a unified end-to-end architecture composed of a lightweight language backbone, a Whisper-based audio encoder, and a sparsely activated Mixture-of-Experts (MoE) adapter that explicitly accounts for audio heterogeneity and alleviates cross-modal optimization conflicts under limited capacity. To further enhance paralinguistic reasoning, we introduce DataFlux, a closed loop audio instruction data synthesis and verification pipeline that constructs high quality, logically consistent supervision from raw audio. Extensive evaluations across ASR, knowledge reasoning, safety, instruction following, and paralinguistic benchmarks, demonstrate that Eureka-Audio achieves an efficient balance between computational cost and performance. These results establish Eureka Audio as a strong and practical baseline for lightweight audio understanding models.

Current open-source diffusion models struggle to generate stable and synchronized audio-visual content, particularly in scenarios demanding complex semantic reasoning. The root cause is that existing methods rely on coarse text embeddings from off-the-shelf encoders to guide audio-video denoising, which discards fine-grained semantics and, critically, lacks a shared long-horizon plan, leading to uncoordinated denoising trajectories and fragile cross-modal alignment. We propose Baton, the first framework that introduces explicit semantic planning into joint video-audio generation. Our key insight is that complementing coarse text guidance with semantically rich, modality-aware planned tokens, jointly reasoned and mutually aligned before denoising, can simultaneously restore fine-grained semantic detail and establish a shared blueprint that coordinates both audio and video denoising trajectories. Concretely, Baton first introduces the VA-Planner, a multimodal language model equipped with dual semantic alignment towers, where learnable queries cross-attend to both video and audio features to produce a pair of semantically aligned video and audio planned tokens as keyframe-level blueprints. These planned tokens are injected into the diffusion backbone via cross-attention layers, providing temporally grounded guidance complementary to coarse text embeddings. Since planned tokens do not share one-to-one spatial-temporal correspondence with diffusion latents, we further propose Relative Semantic RoPE, a relative positional encoding that maps planned tokens and latents into a shared spatial-temporal coordinate frame, enabling each latent to accurately attend to its positionally corresponding semantic cues. Experiments on benchmarks show the effectiveness of Baton both qualitatively and quantitatively.

Speech-to-text (S2T) systems for recognition (ASR) and translation (S2TT) typically generate discrete text tokens. In contrast, continuous-target language modelling performs generation in a continuous space, yet its potential for S2T remains unexplored. To bridge this gap, we propose ELF-S2T, an audio-conditioned continuous-target generative model for S2T. Built upon the pre-trained Embedded Language Flows (ELF) backbone, ELF-S2T processes speech via a frozen Whisper encoder and a single linear projector, prepending the resulting audio condition to the noisy text latent for in-context, flow-matching denoising. To prevent the model from over-relying on its pre-trained text context, we introduce audio forcing during training, and further amplify the audio condition via classifier-free guidance at inference. Experiments on LibriSpeech and CoVoST2 show that ELF-S2T achieves competitive ASR and S2TT performance. Crucially, our error analysis reveals that, although ASR and S2TT errors look very different on the surface, both stem from the same underlying cause, a close distance confusion in the continuous latent space. This finding naturally aligns with the continuous representation generation paradigm, indicating a common semantic mapping process beneath recognition and translation. Our code and pretrained models are publicly available at https://github.com/Sslnon/ELF-S2T.

Automatic speech recognition is dominated by autoregressive decoders that emit one token at a time. We ask whether a discrete diffusion language model can transcribe speech instead, refining a whole transcript in parallel over a small number of denoising steps. We train an audio-native interface for DiffusionGemma, a 26B mixture-of-experts model that generates text by uniform, random-token discrete diffusion rather than the absorbing-mask scheme common to recent diffusion language models. A frozen Whisper encoder supplies acoustic features, a lightweight projector maps them into the model embedding space, and low-rank adapters let the frozen backbone attend to the new modality. About 42M parameters are trained, which is 0.16 percent of the backbone. We find that the natural training objectives fail to ground the audio because their gradient reaches the projector only through attention that has already dismissed it. A connectionist temporal classification loss applied through the frozen output head breaks this deadlock. The resulting model reaches 6.6 percent word error rate on LibriSpeech test-clean, transcribes in roughly eight parallel steps regardless of utterance length, and uses a single adapter trained on six languages, which we evaluate here on English, Hindi, and Mandarin.

In this paper, we present an efficient neural network for end-to-end general purpose audio source separation. Specifically, the backbone structure of this convolutional network is the SUccessive DOwnsampling and Resampling of Multi-Resolution Features (SuDoRMRF) as well as their aggregation which is performed through simple one-dimensional convolutions. In this way, we are able to obtain high quality audio source separation with limited number of floating point operations, memory requirements, number of parameters and latency. Our experiments on both speech and environmental sound separation datasets show that SuDoRMRF performs comparably and even surpasses various state-of-the-art approaches with significantly higher computational resource requirements.

We introduce Qwen-Audio-VAE, a suite of low-bitrate, fast-encoding continuous audio autoencoders designed for scalable general audio generation. The model is built around a simple but important principle: an audio VAE should not only reconstruct diverse audio with high fidelity, but also produce compact latent representations fast enough to support large-scale text-to-audio training. Qwen-Audio-VAE combines a causal encoder-decoder, window Transformer blocks, and multi-discriminator training to achieve a strong balance between reconstruction quality and compression rate. The model is trained at scale on 5 million hours of multi-domain audio, enabling robust reconstruction across heterogeneous acoustic conditions. To further improve computational efficiency, we adopt an asymmetric encoder-decoder backbone and introduce latency-aware encoder pruning to maximize encoding throughput. Experiments on public speech, music, and sound reconstruction benchmarks show that Qwen-Audio-VAE generalizes well across diverse audio domains and is particularly efficient, requiring only 541 ms to encode 32 minutes of audio. Overall, Qwen-Audio-VAE provides a high-quality, compact, and high-throughput representation backbone for efficient general audio generation.

Text-guided audio editing aims to modify the language-specified acoustic content while preserving edit-irrelevant source components. Existing training-free methods typically rely on inversion-based editing. While inversion-free editing is appealing as it decreases computational overhead and reconstruction errors, it remains largely unexplored for audio editing. The key challenge is to construct a source-to-target editing path through diffusion denoising dynamics. In this paper, we introduce DirectAudioEdit, the first attempt to develop a training-free and inversion-free method for audio editing. Experiments on music and event-level benchmarks across two backbones show that DirectAudioEdit reduces macro-averaged FAD and KL by 15.9% and 15.8% compared with DDPM inversion, while achieving up to 64.5% editing speedup.

Streaming audio-visual target speaker extraction (AVTSE) is essential for latency-sensitive applications. However, current causal systems face two coupled gaps. First, standard benchmarks fix the number of speakers at two and ignore the near-/far-field interference, reverberation, and sudden noise of real scenes. Second, high-performance causal architectures are too heavy for edge deployment. Lightweight alternatives try to reclaim capacity by repeatedly invoking a small separator, which offsets the savings. To address the first gap, we release RealSSA, a realistic AVTSE benchmark consisting of RealSSA-Sim and RealSSA-Real. RealSSA-Sim provides controllable simulated scenes with 1--5 dynamic near-field interfering speakers; scenes with far-field interference additionally include 10--15 far-field interfering speakers. RealSSA-Real provides a held-out real-recorded evaluation set. To address the second gap, we propose Falcon, a causal AVTSE method. Its Separator performs multi-scale time-frequency separation in a single encoder--decoder pass. We train Falcon with an encoder-space speaker contrastive loss. This loss suppresses near-field leakage at zero inference cost and transfers across backbones. Falcon achieves state-of-the-art extraction quality on RealSSA-Sim, LRS2-2Mix, LRS3-2Mix, and VoxCeleb2-2Mix. Compared with the causal AV-TFGridNet baseline, it cuts parameters by 91.9%, computation by 20.2 times, and GPU inference latency by 11.1 times. Code and demo are available at https://cslikai.cn/Falcon/.

Automated audio captioning (AAC) aims to generate informative descriptions for various sounds from nature and/or human activities. In recent years, AAC has quickly attracted research interest, with state-of-the-art systems now relying on a sequence-to-sequence (seq2seq) backbone powered by strong models such as Transformers. Following the macro-trend of applied machine learning research, in this work, we strive to improve the performance of seq2seq AAC models by extensively leveraging pretrained models and large language models (LLMs). Specifically, we utilize BEATs to extract fine-grained audio features. Then, we employ Instructor LLM to fetch text embeddings of captions, and infuse their language-modality knowledge into BEATs audio features via an auxiliary InfoNCE loss function. Moreover, we propose a novel data augmentation method that uses ChatGPT to produce caption mix-ups (i.e., grammatical and compact combinations of two captions) which, together with the corresponding audio mixtures, increase not only the amount but also the complexity and diversity of training data. During inference, we propose to employ nucleus sampling and a hybrid reranking algorithm, which has not been explored in AAC research. Combining our efforts, our model achieves a new state-of-the-art 32.6 SPIDEr-FL score on the Clotho evaluation split, and wins the 2023 DCASE AAC challenge.

Large Audio-Language Models (LALMs) have significantly advanced intelligent human-computer interaction, yet their reliance on text-based outputs limits their ability to generate natural speech responses directly, hindering seamless audio interactions. To address this, we introduce Step-Audio-AQAA, a fully end-to-end LALM designed for Audio Query-Audio Answer (AQAA) tasks. The model integrates a dual-codebook audio tokenizer for linguistic and semantic feature extraction, a 130-billion-parameter backbone LLM and a neural vocoder for high-fidelity speech synthesis. Our post-training approach employs interleaved token-output of text and audio to enhance semantic coherence and combines Direct Preference Optimization (DPO) with model merge to improve performance. Evaluations on the StepEval-Audio-360 benchmark demonstrate that Step-Audio-AQAA excels especially in speech control, outperforming the state-of-art LALMs in key areas. This work contributes a promising solution for end-to-end LALMs and highlights the critical role of token-based vocoder in enhancing overall performance for AQAA tasks.

Recent advancements in audio-aware large language models (ALLMs) enable them to process and understand audio inputs. However, these models often hallucinate non-existent sound events, reducing their reliability in real-world applications. To address this, we propose LISTEN (Learning to Identify Sounds Through Extended Negative Samples), a contrastive-like training method that enhances ALLMs' ability to distinguish between present and absent sounds using synthesized data from the backbone LLM. Unlike prior approaches, our method requires no modification to LLM parameters and efficiently integrates audio representations via a lightweight adapter. Experiments show that LISTEN effectively mitigates hallucinations while maintaining impressive performance on existing audio question and reasoning benchmarks. At the same time, it is more efficient in both data and computation.

We introduce DeSTA2.5-Audio, a general-purpose Large Audio Language Model (LALM) designed for robust auditory perception and instruction-following, without requiring task-specific audio instruction-tuning. Recent LALMs typically augment Large Language Models (LLMs) with auditory capabilities by training on large-scale, manually curated or LLM-synthesized audio-instruction datasets. However, these approaches have often suffered from the catastrophic forgetting of the LLM's original language abilities. To address this, we revisit the data construction pipeline and propose DeSTA, a self-generated cross-modal alignment strategy in which the backbone LLM generates its own training targets. This approach preserves the LLM's native language proficiency while establishing effective audio-text alignment, thereby enabling zero-shot generalization without task-specific tuning. Using DeSTA, we construct DeSTA-AQA5M, a large-scale, task-agnostic dataset containing 5 million training samples derived from 7,000 hours of audio spanning 50 diverse datasets, including speech, environmental sounds, and music. DeSTA2.5-Audio achieves state-of-the-art or competitive performance across a wide range of audio-language benchmarks, including Dynamic-SUPERB, MMAU, SAKURA, Speech-IFEval, and VoiceBench. Comprehensive comparative studies demonstrate that our self-generated strategy outperforms widely adopted data construction and training strategies in both auditory perception and instruction-following capabilities. Our findings underscore the importance of carefully designed data construction in LALM development and offer practical insights for building robust, general-purpose LALMs.

The rapid advancement of generative audio models has outpaced the development of robust evaluation methodologies. Existing objective metrics and general multimodal large language models (MLLMs) often struggle with domain generalization, zero-shot capabilities, and instructional flexibility. To address these bottlenecks, we propose JASTIN, a generalizable, instruction-driven audio evaluation framework that formulates audio assessment as a self-instructed reasoning task. JASTIN bridges a frozen high-performance audio encoder with a fine-tuned LLM backbone via a trainable audio adapter. To ensure robust zero-shot generalization, we introduce a comprehensive instruction following data preparation pipeline, incorporating Multi-Source, Multi-Task, Multi-Calibration, and Multi-Description data. Experimental results demonstrate that JASTIN achieves state-of-the-art Pearson and Spearman correlations with human subjective ratings. It consistently outperforms general MLLMs across speech, sound, music, and out-of-domain evaluation tasks without the need for task-specific retraining.

Recent advances in joint audio-video generation have been remarkable, yet real-world applications demand strong per-modality fidelity, cross-modal alignment, and fine-grained synchronization. Reinforcement Learning (RL) offers a promising paradigm, but its extension to multi-objective and multi-modal joint audio-video generation remains unexplored. Notably, our in-depth analysis first reveals that the primary obstacles to applying RL in this stem from: (i) multi-objective advantages inconsistency, where the advantages of multimodal outputs are not always consistent within a group; (ii) multi-modal gradients imbalance, where video-branch gradients leak into shallow audio layers responsible for intra-modal generation; (iii) uniform credit assignment, where fine-grained cross-modal alignment regions fail to get efficient exploration. These shortcomings suggest that vanilla RL fine-tuning strategy with a single global advantage often leads to suboptimal results. To address these challenges, we propose OmniNFT, a novel modality-aware online diffusion RL framework with three key innovations: (1) Modality-wise advantage routing, which routes independent per-reward advantages to their respective modality generation branches. (2) Layer-wise gradient surgery, which selectively detaches video-branch gradients on shallow audio layers while retaining those for cross-modal interaction layers. (3) Region-wise loss reweighting, which modulates policy optimization toward critical regions related to audio-video synchronization and fine-grained alignment. Extensive experiments on JavisBench and VBench with the LTX-2 backbone demonstrate that OmniNFT achieves comprehensive improvements in audio and video perceptual quality, cross-modal alignment, and audio-video synchronization.

Large Audio-Language Models (LALMs) are often constrained by short audio context windows, even when their text backbones support long contexts, limiting long-form audio understanding. Prior work has introduced context-extension methods (e.g. YaRN) on unimodal LLMs, yet their application to LALMs remains unexplored. First, building on RoPE-based context extension, we introduce Partial YaRN, a training-free, audio-only extension method that modifies only audio token positions, leaving text positions intact to preserve the base LLM's text capabilities. Second, we propose Virtual Longform Audio Training (VLAT), a training strategy that extends Partial YaRN into a training-time positional augmentation. VLAT simulates diverse audio lengths during training, enabling generalization to inputs far longer than those seen in training and improving robustness for long-context audio understanding. Our experiments on SALMONN and Qwen2-Audio show that Partial YaRN outperforms the original models across wide range of settings, and VLAT training strategy provides substantial improvement, achieving strong performance on long audio of unseen lengths.

Most deepfake detection methods focus on detecting spatial and/or spatio-temporal changes in facial attributes. This is because available benchmark datasets contain mostly visual-only modifications. However, a sophisticated deepfake may include small segments of audio or audio-visual manipulations that can completely change the meaning of the content. To addresses this gap, we propose and benchmark a new dataset, Localized Audio Visual DeepFake (LAV-DF), consisting of strategic content-driven audio, visual and audio-visual manipulations. The proposed baseline method, Boundary Aware Temporal Forgery Detection (BA-TFD), is a 3D Convolutional Neural Network-based architecture which efficiently captures multimodal manipulations. We further improve (i.e. BA-TFD+) the baseline method by replacing the backbone with a Multiscale Vision Transformer and guide the training process with contrastive, frame classification, boundary matching and multimodal boundary matching loss functions. The quantitative analysis demonstrates the superiority of BA- TFD+ on temporal forgery localization and deepfake detection tasks using several benchmark datasets including our newly proposed dataset. The dataset, models and code are available at https://github.com/ControlNet/LAV-DF.

Speech understanding as an element of the more generic video understanding using audio-visual large language models (av-LLMs) is a crucial yet understudied aspect. This paper proposes video-SALMONN, a single end-to-end av-LLM for video processing, which can understand not only visual frame sequences, audio events and music, but speech as well. To obtain fine-grained temporal information required by speech understanding, while keeping efficient for other video elements, this paper proposes a novel multi-resolution causal Q-Former (MRC Q-Former) structure to connect pre-trained audio-visual encoders and the backbone large language model. Moreover, dedicated training approaches including the diversity loss and the unpaired audio-visual mixed training scheme are proposed to avoid frames or modality dominance. On the introduced speech-audio-visual evaluation benchmark, video-SALMONN achieves more than 25\% absolute accuracy improvements on the video-QA task and over 30\% absolute accuracy improvements on audio-visual QA tasks with human speech. In addition, video-SALMONN demonstrates remarkable video comprehension and reasoning abilities on tasks that are unprecedented by other av-LLMs. Our training code and model checkpoints are available at \url{https://github.com/bytedance/SALMONN/}.

In this paper, we introduce SoloAudio, a novel diffusion-based generative model for target sound extraction (TSE). Our approach trains latent diffusion models on audio, replacing the previous U-Net backbone with a skip-connected Transformer that operates on latent features. SoloAudio supports both audio-oriented and language-oriented TSE by utilizing a CLAP model as the feature extractor for target sounds. Furthermore, SoloAudio leverages synthetic audio generated by state-of-the-art text-to-audio models for training, demonstrating strong generalization to out-of-domain data and unseen sound events. We evaluate this approach on the FSD Kaggle 2018 mixture dataset and real data from AudioSet, where SoloAudio achieves the state-of-the-art results on both in-domain and out-of-domain data, and exhibits impressive zero-shot and few-shot capabilities. Source code and demos are released.

Modeling temporal characteristics plays a significant role in the representation learning of audio waveform. We propose Contrastive Long-form Language-Audio Pretraining (CoLLAP) to significantly extend the perception window for both the input audio (up to 5 minutes) and the language descriptions (exceeding 250 words), while enabling contrastive learning across modalities and temporal dynamics. Leveraging recent Music-LLMs to generate long-form music captions for full-length songs, augmented with musical temporal structures, we collect 51.3K audio-text pairs derived from the large-scale AudioSet training dataset, where the average audio length reaches 288 seconds. We propose a novel contrastive learning architecture that fuses language representations with structured audio representations by segmenting each song into clips and extracting their embeddings. With an attention mechanism, we capture multimodal temporal correlations, allowing the model to automatically weigh and enhance the final fusion score for improved contrastive alignment. Finally, we develop two variants of the CoLLAP model with different types of backbone language models. Through comprehensive experiments on multiple long-form music-text retrieval datasets, we demonstrate consistent performance improvement in retrieval accuracy compared with baselines. We also show the pretrained CoLLAP models can be transferred to various music information retrieval tasks, with heterogeneous long-form multimodal contexts.

We present daVinci-MagiHuman, an open-source audio-video generative foundation model for human-centric generation. daVinci-MagiHuman jointly generates synchronized video and audio using a single-stream Transformer that processes text, video, and audio within a unified token sequence via self-attention only. This single-stream design avoids the complexity of multi-stream or cross-attention architectures while remaining easy to optimize with standard training and inference infrastructure. The model is particularly strong in human-centric scenarios, producing expressive facial performance, natural speech-expression coordination, realistic body motion, and precise audio-video synchronization. It supports multilingual spoken generation across Chinese (Mandarin and Cantonese), English, Japanese, Korean, German, and French. For efficient inference, we combine the single-stream backbone with model distillation, latent-space super-resolution, and a Turbo VAE decoder, enabling generation of a 5-second 256p video in 2 seconds on a single H100 GPU. In automatic evaluation, daVinci-MagiHuman achieves the highest visual quality and text alignment among leading open models, along with the lowest word error rate (14.60%) for speech intelligibility. In pairwise human evaluation, it achieves win rates of 80.0% against Ovi 1.1 and 60.9% against LTX 2.3 over 2000 comparisons. We open-source the complete model stack, including the base model, the distilled model, the super-resolution model, and the inference codebase.

Joint audio-video generation models have shown that unified generation yields stronger cross-modal coherence than cascaded approaches. However, existing models couple modalities throughout denoising via pervasive attention, treating high-level semantics and low-level details in a fully entangled manner. This is suboptimal for talking head synthesis: while audio and facial motion are semantically correlated, their low-level realizations (acoustic signals and visual textures) follow distinct rendering processes. Enforcing joint modeling across all levels causes unnecessary entanglement and reduces efficiency. We propose Talker-T2AV, an autoregressive diffusion framework where high-level cross-modal modeling occurs in a shared backbone, while low-level refinement uses modality-specific decoders. A shared autoregressive language model jointly reasons over audio and video in a unified patch-level token space. Two lightweight diffusion transformer heads decode the hidden states into frame-level audio and video latents. Experiments on talking portrait benchmarks show Talker-T2AV outperforms dual-branch baselines in lip-sync accuracy, video quality, and audio quality, achieving stronger cross-modal consistency than cascaded pipelines.

Audio intelligence involves understanding, reasoning about, and generating both audio and speech. In this work, we introduce Nemotron-Labs-Audex-30B-A3B (Audex), a unified audio-text LLM built on Nemotron-Cascade-2-30B-A3B, a strong text-only MoE LLM. Audex adopts a simple unified design with a single Transformer decoder: audio inputs are encoded and projected into the text embedding space, while text tokens and quantized audio output tokens are treated uniformly during generation. This architecture enables strong audio-text fusion, seamless multimodal generation, and compatibility with standard LLM training and inference infrastructure. For training, we meticulously curate audio-text datasets comprising 157.4B audio tokens and 320.5B text tokens. We apply multi-stage supervised training on these datasets, followed by text-only Cascade RL and multi-domain on-policy distillation. Audex delivers state-of-the-art audio understanding, speech recognition and translation, text-to-speech, audio generation, and speech-to-speech generation, while preserving very compelling reasoning, alignment, knowledge, long-context, and agentic capabilities of its text-only LLM backbone with marginal or no regression. We release the model checkpoints to facilitate open research.

Text and lyrics specify broad musical characteristics and sung content but offer limited control over musical timing, melody, and reference-based style. We introduce DiffSynth-Music (https://modelscope.cn/models/DiffSynth-Studio/DiffSynth-Music), a framework that adds composable audio conditioning to a music synthesis backbone through layer-wise key-value injection. The three template models, Control, Prosody, and Reference, are initialized from the backbone diffusion transformer and trained with conditional flow matching. They support five control types: beats, vocals, accompaniment, prosody, and reference audio. A shared variational autoencoder maps conditioning waveforms into a common latent space, enabling their attention memories to be combined. With the template timestep fixed at the clean-data endpoint and other inputs held constant, each control cache is computed once and reused throughout sampling. Training pairs are derived from music recordings using beat extraction, source separation, vocal resynthesis, and reference-excerpt selection. Single-control evaluations on Mandarin and English songs demonstrate improved adherence across all five control types and better lyric fidelity under vocal conditioning relative to the backbone. Automatic music-quality and instruction-following scores remain broadly comparable to those of the evaluated base models, with metric-specific trade-offs. We release the three template models to support research and creative applications in controllable music generation.

Audio-visual video segmentation~(AVVS) aims to generate pixel-level maps of sound-producing objects within image frames and ensure the maps faithfully adhere to the given audio, such as identifying and segmenting a singing person in a video. However, existing methods exhibit two limitations: 1) they address video temporal features and audio-visual interactive features separately, disregarding the inherent spatial-temporal dependence of combined audio and video, and 2) they inadequately introduce audio constraints and object-level information during the decoding stage, resulting in segmentation outcomes that fail to comply with audio directives. To tackle these issues, we propose a decoupled audio-video transformer that combines audio and video features from their respective temporal and spatial dimensions, capturing their combined dependence. To optimize memory consumption, we design a block, which, when stacked, enables capturing audio-visual fine-grained combinatorial-dependence in a memory-efficient manner. Additionally, we introduce audio-constrained queries during the decoding phase. These queries contain rich object-level information, ensuring the decoded mask adheres to the sounds. Experimental results confirm our approach's effectiveness, with our framework achieving a new SOTA performance on all three datasets using two backbones. The code is available at https://github.com/aspirinone/CATR.github.io

Deep learning approaches for black-box modelling of audio effects have shown promise, however, the majority of existing work focuses on nonlinear effects with behaviour on relatively short time-scales, such as guitar amplifiers and distortion. While recurrent and convolutional architectures can theoretically be extended to capture behaviour at longer time scales, we show that simply scaling the width, depth, or dilation factor of existing architectures does not result in satisfactory performance when modelling audio effects such as fuzz and dynamic range compression. To address this, we propose the integration of time-varying feature-wise linear modulation into existing temporal convolutional backbones, an approach that enables learnable adaptation of the intermediate activations. We demonstrate that our approach more accurately captures long-range dependencies for a range of fuzz and compressor implementations across both time and frequency domain metrics. We provide sound examples, source code, and pretrained models to faciliate reproducibility.

Text-to-video (T2V) diffusion models have recently achieved impressive visual quality, yet most systems still generate silent clips and treat audio as a secondary concern. Existing audio-video generation pipelines typically decompose the task into cascaded stages, which accumulate errors across modalities and are trained under separate objectives. Recent joint audio-video generators alleviate this issue but often rely on dual-tower architectures with ad-hoc cross-modal bridges and static, single-shot text conditioning, making it difficult to both reuse T2V backbones and to reason about how audio, video and language interact over time. To address these challenges, we propose 3MDiT, a unified tri-modal diffusion transformer for text-driven synchronized audio-video generation. Our framework models video, audio and text as jointly evolving streams: an isomorphic audio branch mirrors a T2V backbone, tri-modal omni-blocks perform feature-level fusion across the three modalities, and an optional dynamic text conditioning mechanism updates the text representation as audio and video evidence co-evolve. The design supports two regimes: training from scratch on audio-video data, and orthogonally adapting a pretrained T2V model without modifying its backbone. Experiments show that our approach generates high-quality videos and realistic audio while consistently improving audio-video synchronization and tri-modal alignment across a range of quantitative metrics.

Audio-Visual Target Speaker Extraction (AV-TSE) aims to mimic the human ability to enhance auditory perception using visual cues. Although numerous models have been proposed recently, most of them estimate target signals by primarily relying on local dependencies within acoustic features, underutilizing the human-like capacity to infer unclear parts of speech through contextual information. This limitation results in not only suboptimal performance but also inconsistent extraction quality across the utterance, with some segments exhibiting poor quality or inadequate suppression of interfering speakers. To close this gap, we propose a model-agnostic strategy called the Mask-And-Recover (MAR). It integrates both inter- and intra-modality contextual correlations to enable global inference within extraction modules. Additionally, to better target challenging parts within each sample, we introduce a Fine-grained Confidence Score (FCS) model to assess extraction quality and guide extraction modules to emphasize improvement on low-quality segments. To validate the effectiveness of our proposed model-agnostic training paradigm, six popular AV-TSE backbones were adopted for evaluation on the VoxCeleb2 dataset, demonstrating consistent performance improvements across various metrics.

Handling fast, complex, and highly non-linear motion patterns has long posed challenges for video frame interpolation. Although recent diffusion-based approaches improve upon traditional optical-flow-based methods, they still struggle to cover diverse application scenarios and often fail to produce sharp, temporally consistent frames in fine-grained motion tasks such as audio-visual synchronized interpolation. To address these limitations, we introduce BBF (Beyond Boundary Frames), a context-aware video frame interpolation framework, which could be guided by audio/visual semantics. First, we enhance the input design of the interpolation model so that it can flexibly handle multiple conditional modalities, including text, audio, images, and video. Second, we propose a decoupled multimodal fusion mechanism that sequentially injects different conditional signals into a DiT backbone. Finally, to maintain the generation abilities of the foundation model, we adopt a progressive multi-stage training paradigm, where the start-end frame difference embedding is used to dynamically adjust both the data sampling and the loss weighting. Extensive experimental results demonstrate that BBF outperforms specialized state-of-the-art methods on both generic interpolation and audio-visual synchronized interpolation tasks, establishing a unified framework for video frame interpolation under coordinated multi-channel conditioning.

Audio generation and audio-to-text understanding remain largely separate, with diffusion models dominating high-fidelity synthesis and autoregressive (AR) language models driving captioning and semantic prediction. Existing unified approaches typically rely on either heterogeneous modules or AR-centric modeling, which can hinder joint optimization and limit acoustic fidelity. We present UAT, to our knowledge, the first diffusion-centric framework that supports unified audio generation, editing, and captioning. UAT couples continuous latent diffusion for audio with masked discrete diffusion for text, enabling bidirectional audio-text modeling within a shared dual-stream backbone. Experiments show that UAT preserves strong audio generation and editing capabilities while achieving competitive captioning performance, demonstrating a favorable balance between acoustic synthesis and semantic prediction. Demo samples are available at https://UAT-demo.github.io.

A single embedding space that covers text, images, video, and audio lets one index serve every query a user can pose. Embedding models built on vision-language backbones now lead text/image/video retrieval benchmarks but lack audio entirely, while audio-text retrieval is led by specialist systems that serve no other modality. We present the Fusion Embedding family, which adds audio to a frozen vision-language embedding base whose parameters are never updated: generation 1 (fusion-embedding-1) trains only a 16.4M-parameter connector between a frozen audio tower and the frozen base, and generation 2 (fusion-embedding-2) adds modality-gated deep adapters (44.2M parameters) whose branch never executes on text, image, or video inputs: their outputs are bit-for-bit those of the released base, verified after every training run. Because the base already binds text, images, and video, aligning audio to text alone makes audio-image retrieval emerge, with zero paired audio-visual training data. Alongside the recipe we map its design space with controlled negative results (rewriting training captions with an LLM, substituting a leaderboard-stronger audio tower, and widening the connector each reduce retrieval) and with training-protocol findings that we expect to transfer to any frozen decoder-LM embedding backbone. Both generations train in hours on a single GPU. Weights, code, and the evaluation harness are openly released.

Developing architectures suitable for modeling raw audio is a challenging problem due to the high sampling rates of audio waveforms. Standard sequence modeling approaches like RNNs and CNNs have previously been tailored to fit the demands of audio, but the resultant architectures make undesirable computational tradeoffs and struggle to model waveforms effectively. We propose SaShiMi, a new multi-scale architecture for waveform modeling built around the recently introduced S4 model for long sequence modeling. We identify that S4 can be unstable during autoregressive generation, and provide a simple improvement to its parameterization by drawing connections to Hurwitz matrices. SaShiMi yields state-of-the-art performance for unconditional waveform generation in the autoregressive setting. Additionally, SaShiMi improves non-autoregressive generation performance when used as the backbone architecture for a diffusion model. Compared to prior architectures in the autoregressive generation setting, SaShiMi generates piano and speech waveforms which humans find more musical and coherent respectively, e.g. 2x better mean opinion scores than WaveNet on an unconditional speech generation task. On a music generation task, SaShiMi outperforms WaveNet on density estimation and speed at both training and inference even when using 3x fewer parameters. Code can be found at https://github.com/HazyResearch/state-spaces and samples at https://hazyresearch.stanford.edu/sashimi-examples.

We propose AV-Link, a unified framework for Video-to-Audio and Audio-to-Video generation that leverages the activations of frozen video and audio diffusion models for temporally-aligned cross-modal conditioning. The key to our framework is a Fusion Block that enables bidirectional information exchange between our backbone video and audio diffusion models through a temporally-aligned self attention operation. Unlike prior work that uses feature extractors pretrained for other tasks for the conditioning signal, AV-Link can directly leverage features obtained by the complementary modality in a single framework i.e. video features to generate audio, or audio features to generate video. We extensively evaluate our design choices and demonstrate the ability of our method to achieve synchronized and high-quality audiovisual content, showcasing its potential for applications in immersive media generation. Project Page: snap-research.github.io/AVLink/

This paper proposes an architecture for equipping large language models (LLMs) with audio-understanding capabilities without fine-tuning their weights. The proposed symbiotic architecture employs an injector module that writes audio-conditioned vectors directly into the target LLM's short-term memory, i.e., the key-value (KV) cache, enabling the LLM to behave as an audio language model (ALM). The architectural advantages are twofold. First, it improves the scalability of ALMs: because the proposed method bypasses the LLM during audio injection, the injection cost is governed by the injector width rather than the backbone width, and can therefore scale more slowly than the cost of full-backbone prefilling. Second, since the training scheme does not update the LLM weights, the original capabilities of the LLM are preserved without the risk of degradation from fine-tuning. The effectiveness of the proposed method is evaluated on both audio-understanding tasks (automatic speech recognition, audio question answering, and acoustic scene classification) and text-only tasks. We confirm that, while activating fewer parameters during audio prefilling, our architecture outperforms the conventional method with a frozen LLM and approaches the performance of a fine-tuned ALM, all while preserving the backbone LLM's original text-only task performance by construction.

Audio language models can understand audio inputs and perform a range of audio-related tasks based on instructions, such as speech recognition and audio captioning, where the instructions are usually textual prompts. Audio language models are mostly initialized from pre-trained audio encoders and large language models (LLMs). Although these pre-trained components were developed to support multiple languages, audio-language models are trained predominantly on English data, which may limit their usability to only English instructions or English speech inputs. First, this paper examines the performance of existing audio language models in an underserved language using Thai as an example. This paper demonstrates that, despite being built on multilingual backbones, audio language models do not exhibit cross-lingual emergent abilities to low-resource languages. Second, this paper studies data mixture for developing audio language models that are optimized for a target language as well as English. In addition. this paper integrates audio comprehension and speech instruction-following capabilities into a single unified model. Our experiments provide insights into data mixture for enhancing instruction-following capabilities in both a low-resource language and English. Our model, Typhoon-Audio, outperforms existing open-source audio language models by a considerable margin, and it is comparable to state-of-the-art Gemini-1.5-Pro in both English and Thai languages.

Generating a coherent multi-shot video requires structured cross-shot memory. Subject appearance, scene context, and speaker identity must persist across cuts. Existing approaches either train end-to-end over fixed-length sequences and cannot scale, generate shot-by-shot with memory banks that grow linearly, or orchestrate pretrained generators under an LLM planner without a multi-shot-aware backbone. We present UnityShots, a memory-driven multi-shot audio-video generation system built on LTX-2.3, trained on annotated cinematic and music-video shots. The video stream maintains two fixed-size slots, a long-term memory (LTM) slot anchored to the opening shot and a short-term memory (STM) slot holding the immediately preceding tail, both updated at every cut by a boundary-conditioned gate that fuses visual cut probability and beat-tracker signals. The audio stream injects a reference speaker token at every shot to preserve vocal timbre without a sliding audio bank. A discrete cut-type prior, learned through AdaLN, becomes an inference-time control knob over transition strength. We release a benchmark of 200 multi-cultural multi-shot sequences spanning six ethnic regions and ten or more languages, with per-shot reference identities, reference audio, and per-boundary transition labels. Evaluated across I2V, T2V, and R2V conditioning modes, UnityShots leads open-source baselines on every cross-shot coherence metric and matches the strongest closed-source system on the multi-shot axes.

As multimodal content continues to expand at a rapid pace, audio retrieval has emerged as a key enabling technology for media search, content organization, and intelligent assistants. However, most existing benchmarks concentrate on semantic matching and fail to capture the fact that real-world queries often demand advanced reasoning abilities, including negation understanding, temporal ordering, concurrent event recognition, and duration discrimination. To address this gap, we introduce ReasonAudio, the first reasoning-intensive benchmark for Text-Audio Retrieval, comprising 1,000 queries and 10,000 composite audio clips across five fundamental reasoning tasks: Negation, Order, Overlap, Duration, and Mix. Despite their intuitive nature for humans and straightforward construction, these tasks pose significant challenges to current models. Our evaluation of ten state-of-the-art models reveals the following findings: All models struggle with reasoning-intensive audio retrieval, performing particularly poorly on Negation and Duration while showing relatively better results on Overlap and Order. Moreover, Multimodal Large Language Model-based embedding models fail to inherit the reasoning capabilities of their backbones through contrastive fine-tuning, suggesting that current training paradigms are insufficient to preserve reasoning capacity in retrieval settings

Audio source separation models must preserve fine spectral structure while keeping the cost of the separator tractable. Band-split architectures reduce this cost by manually partitioning the spectrum into predefined frequency bands and encoding each band as a token, but the same boundaries are used for every mixture. Because informative spectral structure varies with source content, fixed boundaries can merge discriminative components or spend tokens where little resolution is needed. We introduce Dynamic Band Splitting (DBS), an end-to-end framework that adapts boundary locations and band widths to each mixture under a fixed band-token budget. DBS predicts content-conditioned boundary scores within frequency blocks, selects a prescribed number of boundaries using hard Top-K, and applies content-dependent pooling to obtain compact band tokens. After separation, it smooths and expands these tokens to the original STFT grid, refines them along frequency, and decodes complex masks with a frequency-shared head. DBS improved separation over the corresponding static models on DnR v2, MUSDB18-HQ, and EchoSet, with DBS-BSRNN achieving relative gain of 29.0% in average SI-SDR on DnR v2. These results and boundary analyses on DnR v2 show that content-adaptive frequency partitioning benefits multiple separator backbones under a fixed band-token budget. Code and audio demos are available at https://cslikai.cn/Dynamic-Band-Splitting

In this paper, we focus on Whisper, a recent automatic speech recognition model trained with a massive 680k hour labeled speech corpus recorded in diverse conditions. We first show an interesting finding that while Whisper is very robust against real-world background sounds (e.g., music), its audio representation is actually not noise-invariant, but is instead highly correlated to non-speech sounds, indicating that Whisper recognizes speech conditioned on the noise type. With this finding, we build a unified audio tagging and speech recognition model Whisper-AT by freezing the backbone of Whisper, and training a lightweight audio tagging model on top of it. With <1% extra computational cost, Whisper-AT can recognize audio events, in addition to spoken text, in a single forward pass.

We present a framework for learning multimodal representations from unlabeled data using convolution-free Transformer architectures. Specifically, our Video-Audio-Text Transformer (VATT) takes raw signals as inputs and extracts multimodal representations that are rich enough to benefit a variety of downstream tasks. We train VATT end-to-end from scratch using multimodal contrastive losses and evaluate its performance by the downstream tasks of video action recognition, audio event classification, image classification, and text-to-video retrieval. Furthermore, we study a modality-agnostic, single-backbone Transformer by sharing weights among the three modalities. We show that the convolution-free VATT outperforms state-of-the-art ConvNet-based architectures in the downstream tasks. Especially, VATT's vision Transformer achieves the top-1 accuracy of 82.1% on Kinetics-400, 83.6% on Kinetics-600, 72.7% on Kinetics-700, and 41.1% on Moments in Time, new records while avoiding supervised pre-training. Transferring to image classification leads to 78.7% top-1 accuracy on ImageNet compared to 64.7% by training the same Transformer from scratch, showing the generalizability of our model despite the domain gap between videos and images. VATT's audio Transformer also sets a new record on waveform-based audio event recognition by achieving the mAP of 39.4% on AudioSet without any supervised pre-training. VATT's source code is publicly available.

Autoregressive (AR) large audio language models (LALMs) such as Qwen-2.5-Omni have achieved strong performance on audio understanding and interaction, but scaling them remains costly in data and computation, and strictly sequential decoding limits inference efficiency. Diffusion large language models (dLLMs) have recently been shown to make effective use of limited training data, and prior work on DIFFA indicates that replacing an AR backbone with a diffusion counterpart can substantially improve audio understanding under matched settings, albeit at a proof-of-concept scale without large-scale instruction tuning, preference alignment, or practical decoding schemes. We introduce DIFFA-2, a practical diffusion-based LALM for general audio understanding. DIFFA-2 upgrades the speech encoder, employs dual semantic and acoustic adapters, and is trained with a four-stage curriculum that combines semantic and acoustic alignment, large-scale supervised fine-tuning, and variance-reduced preference optimization, using only fully open-source corpora. Experiments on MMSU, MMAU, and MMAR show that DIFFA-2 consistently improves over DIFFA and is competitive to strong AR LALMs under practical training budgets, supporting diffusion-based modeling is a viable backbone for large-scale audio understanding. Our code is available at https://github.com/NKU-HLT/DIFFA.git.

Existing audio forgery detection and localization (AFDL) methods often overfit dataset-specific low-level artifacts, limiting their generalization to subtle, localized, and unseen manipulations. Recent audio large language model (ALLM)-based approaches cast AFDL as question answering but still model forensic evidence implicitly, without linking manipulation cues to predictions. To bridge this gap, we propose ThinkOmni, a reasoning-driven omni-modal large language model that jointly performs explicit forensic reasoning, spoofing detection, and temporal manipulation localization. To enable explicit reasoning supervision, we construct Forensic-Aware Chain-of-Thought (FACoT), a 100K-sample dataset with structured forensic evidence and reasoning annotations. Leveraging FACoT, we introduce Forensic-Aware Modality-Incremental Learning (FMIL), which progressively aligns semantic, acoustic, and spectral-visual representations with the LLM backbone to capture complementary forensic cues. We further propose Forensic-Consistent Multi-task Loss (FCML), which combines weighted cross-entropy with an adaptive localization loss to coordinate reasoning generation, spoofing detection, and temporal localization. Extensive experiments show that ThinkOmni achieves strong cross-dataset generalization in both detection and localization. Code, models, data, and inference examples are available at https://beyond0814.github.io/ThinkOmni/.

A common approach to adding audio to a vision-language model is to train or adapt a large omni-modal system. We show that a lightweight alternative can be highly effective for music audio-visual question answering (AVQA). Qwen-MusicAVQA-7B connects a frozen Whisper encoder to Qwen2-VL-7B-Instruct through learned linear projections. The same frozen encoder processes both the video's music track and a TTS-spoken question through separate projectors, while the language model fuses visual frames, music, and question audio through pretrained self-attention, with no task-specific fusion network. On MUSIC-AVQA, our system reaches 96.0% +/- 3.9% accuracy across three independent training seeds on the 7,402-question available-video test subset. Our central finding is that downstream accuracy tracks how much fine-grained local temporal information the audio representation preserves. In a matched 32-token comparison, a stride-pooled Whisper frame sequence outperforms a globally pooled PANNs representation expanded to the same budget by 26 percentage points, even though PANNs sees at least as much audio and uses a far larger projector. The effect is not simply sequence versus vector: within Whisper alone, reducing temporal resolution at a fixed token budget costs a comparable amount. Under matched data and inputs, fine-tuned Qwen2.5-Omni-7B reaches 80.9%, against 95.9% for our 30 s variant; because the systems differ in backbone and adaptation, this is a system-level comparison. Accuracy remains high on sampled head and tail splits of the rephrased MUSIC-AVQA-R benchmark (96.5% and 95.6%). Because both encoders stay frozen and the music features are cached, the entire adaptation is cheap to train: the complete two-stage AVQA run takes approximately 5 hours on a single A100 80GB, and every run reported here fits on that one GPU.

Hour-scale audio-visual question answering is constrained by a context dilemma: dense whole-recording encoding rapidly exhausts context limits, whereas uniform temporal compression severely dilutes fine-grained acoustic and visual evidence. We introduce LEAP, a framework where the model retrieves its own evidence without placing the whole recording in one context. LEAP divides a recording into fixed-duration blocks, applying a lightweight localization pass to each block to score short candidate windows. The highest-ranked windows are pooled and re-encoded in a single bounded answer pass. Consequently, the answer input and peak context remain independent of the recording duration. By decoupling evidence localization from reasoning, our framework can localize candidate temporal windows over pre-computed transcripts without decoding media frames, while preserving fine-grained visual and non-speech evidence by routing the final answering pass over raw audio-visual streams. LEAP trains both stages: a localization LoRA improves the selected windows, and an answer LoRA improves the answers read from the same windows. The block grid natively supports causal queries, enabling LEAP to support streaming inference without streaming-specific training. Across several AVQA benchmarks, LEAP improves over the Qwen3-Omni-30B-A3B baseline by 4.5-16.8%, and transfers to a second omni-modal backbone, MiniCPM-o 4.5, surpassing its published results by 3.1-13.0%.

Large audio-language models (LALMs) demonstrate growing music-understanding capabilities, but whether their responses are grounded in acoustic evidence remains unclear. Musical language often involves abstract concepts whose acoustic evidence is difficult to define and evaluate precisely. We introduce MusicGroundingBench, a controlled benchmark of algorithmically generated piano audio with exact symbolic alignment, comprising three-note and two-bar settings. We evaluate two complementary capabilities: grounding, which localizes the acoustic evidence for a musical query, and understanding, which answers questions about the same excerpts. Our experiments show that cross-modal fine-tuning enables models to learn each capability, but adding grounding supervision does not consistently improve understanding across backbones. We further test whether understanding requires listening through audio-ablation controls that remove or replace the input audio, and use attention analysis to examine whether grounding supervision shifts attention toward note boundaries. Meanwhile, the two evaluated LALMs show limited zero-shot grounding even for basic musical concepts, highlighting grounded music understanding as an important open challenge.

This paper details the DS@GT ARC team's approach to BirdCLEF+ 2026, multi-label detection of animal vocalizations in soundscapes from the Pantanal wetlands. The 2026 edition adds about an hour of labeled soundscapes, shifting the task toward supervised pipelines fit to the labeled set. First, we build a competitive supervised baseline that ensembles a frozen Perch v2 backbone, a trained HGNetV2-B0 sound-event-detection network, and a non-bird prototypical head, reaching a private leaderboard score of 0.936 at rank 1894 within a 90-minute CPU budget. Second, we ask whether token-based representations can compete, contrasting codec representations from neural audio codecs against semantic representations from foundational embeddings. We compare two bioacoustic specialist models against four token-based encoders trained on AudioSet. The repository for this work can be found at https://github.com/dsgt-arc/birdclef-2026.

Strong speech-to-text (S2T) LLMs already provide robust speech perception and text reasoning, but adding speech-to-speech (S2S) output is challenging: fine-tuning the backbone can degrade the original S2T performance, while attaching a downstream talker reintroduces a serial text-to-speech bottleneck. We present PRIME-Speech, a frozen-backbone S2S conversion framework that trains only speech-generation modules. PRIME-Speech synchronizes a causal audio post-decoder with intermediate hidden states of the frozen backbone, so codec tokens are generated from the model's evolving reasoning trajectory rather than from completed text chunks. The post-decoder uses mixed hidden-state, text, and audio-history conditioning, and a training-time packing strategy with turn-level audio KV-cache and position reset stabilizes multi-turn spoken interaction without additional multi-turn S2S training data. Multi-token prediction further reduces the effective codec prediction rate and improves first-audio latency without modifying the reasoning path. Across speech translation, spoken QA, speech understanding, and multi-turn dialogue, PRIME-Speech preserves the S2T behavior of the frozen backbone while producing accurate, low-WER spoken responses.

While large-scale omni-models have demonstrated impressive capabilities across various modalities, their strong performance heavily relies on massive multimodal data and incurs substantial computational costs. This work introduces Speech-Omni-Lite, a cost-efficient framework for extending pre-trained Visual-Language (VL) backbones with speech understanding and generation capabilities, while fully preserving the backbones' vision-language performance. Specifically, the VL backbone is equipped with two lightweight, trainable plug-and-play modules, a speech projector and a speech token generator, while keeping the VL backbone fully frozen. To mitigate the scarcity of spoken QA corpora, a low-cost data construction strategy is proposed to generate Question-Text Answer-Text-Speech (QTATS) data from existing ASR speech-text pairs, facilitating effective speech generation training. Experimental results show that, even with only thousands of hours of speech training data, Speech-Omni-Lite achieves excellent spoken QA performance, which is comparable to omni-models trained on millions of hours of speech data. Furthermore, the learned speech modules exhibit strong transferability across VL backbones.

Streaming speech enhancement requires balancing algorithmic latency against quality, yet existing approaches largely treat this as a binary causal versus non-causal choice. LaCo-SENet addresses this issue with two mechanisms parameterized by a single training-time hyperparameter. First, asymmetric temporal padding redistributes past and future context in convolutions, enabling systematic latency configuration. Second, dual-buffer streaming combines state buffers for past context with lookahead buffers that supply future context at both the input and feature levels. Selective state updates also prevent future-frame leakage into the streaming state, ensuring training-inference consistency. On VoiceBank+DEMAND, a fixed-budget (1.37M parameters) backbone yields a family of models spanning 12.5-75.0 ms, with PESQ rising from 3.35 to 3.43. At just 12.5 ms (fully causal), a PESQ of 3.35 matches or exceeds the prior causal state-of-the-art (3.27 at 46.5 ms).

We present GEPARD (Generative, Prosody-aware, Autoregressive text-to-speech model for Realtime Dialogue), a streaming text-to-speech model for real-time spoken dialogue. GEPARD generates speech autoregressively with an LLM backbone - text and audio embeddings are trained together in a single decoder-only model - and decodes it to a waveform with an FSQ-based neural codec, streaming audio chunk-by-chunk as text arrives. Our central goal is a TTS architecture served by a standard LLM engine (vLLM) without modifying its compute kernels. This defines the overarching design principle: the backbone is a standard full-attention transformer, while all non-trivial auxiliary mechanisms - zero-shot voice cloning, text augmentation, and classifier-free guidance - are moved out of the autoregressive decode loop into prefill, or distilled directly into the weights. On streaming end-to-end inference, a single stream reaches a Real-Time Factor of about 0.067 (roughly 15x faster than real-time); under 256 concurrent streams the system reaches an aggregate speedup of about 204x on a single server-class GPU. We detail: (1) system-level solutions for vLLM-native serving; (2) the "short register" (1-2 word) failure mode of autoregressive speech decoders, with diagnostic probes and a mitigation; and (3) distillation of two-pass classifier-free guidance over text into single-pass weights via Direct Preference Optimization (DPO).

Latent representations are at the heart of the majority of modern generative models. In the audio domain they are typically produced by a neural-audio-codec autoencoder. In this work we introduce SAME (Semantically-Aligned Music autoEncoder), an autoencoder for stereo music and general audio that reaches a 4096times temporal compression ratio while maintaining reconstruction quality and downstream generative performance. We achieve this by combining a tranformer-based backbone with set of semantic regularisation approaches, phase-aware reconstruction losses and improved discriminator designs. The architecture delivers substantial computational cost benefits, through both its high compression ratio and its reliance on well-optimised transformer primitives. Two variants (a large SAME-L and a CPU-deployable SAME-S) are released in open-weights form.

Continuous-latent audio autoencoders form the backbone of latent music generators, yet decoders at high compression rates commonly exhibit three failure modes: high-frequency loss, phase incoherence, and stereo-image collapse. These share a structural root: waveform autoencoders lack an explicit frequency axis, leaving no handle for targeted per-band correction. Among five matched-budget representations, the complex STFT achieves the lowest full-band and high-frequency spectral distances, providing direct access to magnitude and phase at every bin. Building on this, we present ear-VAE2, a complex-spectral autoencoder with cross-channel interaction. Spec-SnakeBeta learns a periodic activation per frequency bin with frequency-dependent initialization, outperforming other activation variants while using fewer parameters than the fully independent variant. Duplex-Aware Refiner applies band-specific corrections to magnitude and phase following duplex theory of sound localization. On the 546-track Song Describer Dataset, ear-VAE2 achieves the best point estimates on five of seven reconstruction metrics. The Duplex-Aware Refiner reduces Mel Distance by 19.4% and uses ~45% fewer residual-output dimensions than the Unconstrained Refiner, while also lowering spectral distances, spatial-cue errors, and receiving higher ratings from professional engineers. The downstream generator using ear-VAE2 latents achieves better point estimates on all 12 automatic metrics.Demo page is available at https://eps-acoustic-revolution-lab.github.io/EAR_VAE2/.

We present Kimi-Audio, an open-source audio foundation model that excels in audio understanding, generation, and conversation. We detail the practices in building Kimi-Audio, including model architecture, data curation, training recipe, inference deployment, and evaluation. Specifically, we leverage a 12.5Hz audio tokenizer, design a novel LLM-based architecture with continuous features as input and discrete tokens as output, and develop a chunk-wise streaming detokenizer based on flow matching. We curate a pre-training dataset that consists of more than 13 million hours of audio data covering a wide range of modalities including speech, sound, and music, and build a pipeline to construct high-quality and diverse post-training data. Initialized from a pre-trained LLM, Kimi-Audio is continual pre-trained on both audio and text data with several carefully designed tasks, and then fine-tuned to support a diverse of audio-related tasks. Extensive evaluation shows that Kimi-Audio achieves state-of-the-art performance on a range of audio benchmarks including speech recognition, audio understanding, audio question answering, and speech conversation. We release the codes, model checkpoints, as well as the evaluation toolkits in https://github.com/MoonshotAI/Kimi-Audio.

Recent advancements in Speech Large Language Models have significantly enhanced multi-dimensional speech understanding. However, the majority of high-performance frameworks are predominantly optimized for GPU centric ecosystems and proprietary backbones, creating a significant gap for deployment on non-CUDA computing infrastructures. In this paper, we present OSUM-Pangu, a fully open-source speech understanding foundation model developed on a completely non-CUDA software and hardware stack. By integrating an audio encoder with the openPangu-7B LLM backbone, we successfully implement the entire training and inference pipeline on the Ascend NPU platform. To facilitate efficient task alignment under non-CUDA resource constraints, we adopt a practical training process that sequentially bridges speech perception and user intent recognition. Experimental results demonstrate that OSUM-Pangu achieves task accuracy comparable to mainstream GPU-based models while maintaining robust natural language interaction capabilities. Our work provides a reproducible, non-CUDA baseline for the open-source speech community, promoting the independent evolution of multimodal intelligence.

Recent advances in diffusion-based lip-syncing generative models have demonstrated their ability to produce highly synchronized talking face videos for visual dubbing. Although these models excel at lip synchronization, they often struggle to maintain fine-grained control over facial details in generated images. In this work, we identify "lip averaging" phenomenon where the model fails to preserve subtle facial details when dubbing unseen in-the-wild videos. This issue arises because the commonly used UNet backbone primarily integrates audio features into visual representations in the latent space via cross-attention mechanisms and multi-scale fusion, but it struggles to retain fine-grained lip details in the generated faces. To address this issue, we propose UnAvgLip, which extracts identity embeddings from reference videos to generate highly faithful facial sequences while maintaining accurate lip synchronization. Specifically, our method comprises two primary components: (1) an Identity Perceiver module that encodes facial embeddings to align with conditioned audio features; and (2) an ID-CrossAttn module that injects facial embeddings into the generation process, enhancing model's capability of identity retention. Extensive experiments demonstrate that, at a modest training and inference cost, UnAvgLip effectively mitigates the "averaging" phenomenon in lip inpainting, significantly preserving unique facial characteristics while maintaining precise lip synchronization. Compared with the original approach, our method demonstrates significant improvements of 5% on the identity consistency metric and 2% on the SSIM metric across two benchmark datasets (HDTF and LRW).

Instruction-guided music editors typically process each request independently, limiting their ability to support workflows in which users progressively refine a track. We introduce AURA, a unified multimodal framework for conversational music editing. AURA uses a multimodal large language model to interpret the complete dialogue history, an optional image, and reference audio, distilling the editing intent into compact concept tokens. A concept-to-audio module injects these tokens and frame-aligned reference features into a frozen MusicGen backbone, enabling precise edits while preserving unaffected content. AURA optimizes only 91M parameters while retaining 1.9B frozen backbone parameters. Experiments on Slakh2100 and MoisesDB demonstrate substantial improvements in edit correctness and content preservation over existing instruction-guided methods, including a 4-5 times reduction in FAD for out-of-domain addition and removal.

Foundation models, now powering most of the exciting applications in deep learning, are almost universally based on the Transformer architecture and its core attention module. Many subquadratic-time architectures such as linear attention, gated convolution and recurrent models, and structured state space models (SSMs) have been developed to address Transformers' computational inefficiency on long sequences, but they have not performed as well as attention on important modalities such as language. We identify that a key weakness of such models is their inability to perform content-based reasoning, and make several improvements. First, simply letting the SSM parameters be functions of the input addresses their weakness with discrete modalities, allowing the model to selectively propagate or forget information along the sequence length dimension depending on the current token. Second, even though this change prevents the use of efficient convolutions, we design a hardware-aware parallel algorithm in recurrent mode. We integrate these selective SSMs into a simplified end-to-end neural network architecture without attention or even MLP blocks (Mamba). Mamba enjoys fast inference (5times higher throughput than Transformers) and linear scaling in sequence length, and its performance improves on real data up to million-length sequences. As a general sequence model backbone, Mamba achieves state-of-the-art performance across several modalities such as language, audio, and genomics. On language modeling, our Mamba-3B model outperforms Transformers of the same size and matches Transformers twice its size, both in pretraining and downstream evaluation.

Learning from multiple modalities, such as audio and video, offers opportunities for leveraging complementary information, enhancing robustness, and improving contextual understanding and performance. However, combining such modalities presents challenges, especially when modalities differ in data structure, predictive contribution, and the complexity of their learning processes. It has been observed that one modality can potentially dominate the learning process, hindering the effective utilization of information from other modalities and leading to sub-optimal model performance. To address this issue the vast majority of previous works suggest to assess the unimodal contributions and dynamically adjust the training to equalize them. We improve upon previous work by introducing a multi-loss objective and further refining the balancing process, allowing it to dynamically adjust the learning pace of each modality in both directions, acceleration and deceleration, with the ability to phase out balancing effects upon convergence. We achieve superior results across three audio-video datasets: on CREMA-D, models with ResNet backbone encoders surpass the previous best by 1.9% to 12.4%, and Conformer backbone models deliver improvements ranging from 2.8% to 14.1% across different fusion methods. On AVE, improvements range from 2.7% to 7.7%, while on UCF101, gains reach up to 6.1%.

We propose a multimodal-driven framework for high-fidelity long-term digital human animation termed Soul, which generates semantically coherent videos from a single-frame portrait image, text prompts, and audio, achieving precise lip synchronization, vivid facial expressions, and robust identity preservation. We construct Soul-1M, containing 1 million finely annotated samples with a precise automated annotation pipeline (covering portrait, upper-body, full-body, and multi-person scenes) to mitigate data scarcity, and we carefully curate Soul-Bench for comprehensive and fair evaluation of audio-/text-guided animation methods. The model is built on the Wan2.2-5B backbone, integrating audio-injection layers and multiple training strategies together with threshold-aware codebook replacement to ensure long-term generation consistency. Meanwhile, step/CFG distillation and a lightweight VAE are used to optimize inference efficiency, achieving an 11.4times speedup with negligible quality loss. Extensive experiments show that Soul significantly outperforms current leading open-source and commercial models on video quality, video-text alignment, identity preservation, and lip-synchronization accuracy, demonstrating broad applicability in real-world scenarios such as virtual anchors and film production. Project page at https://zhangzjn.github.io/projects/Soul/

End-to-end full-duplex speech models feed user audio through an always-on LLM backbone, yet the speaker privacy implications of their hidden representations remain unexamined. Following the VoicePrivacy 2024 protocol with a lazy-informed attacker, we show that the hidden states of SALM-Duplex and Moshi leak substantial speaker identity across all transformer layers. Layer-wise and turn-wise analyses reveal that leakage persists across all layers, with SALM-Duplex showing stronger leakage in early layers while Moshi leaks uniformly, and that Linkability rises sharply within the first few turns. We propose two streaming anonymization setups using Stream-Voice-Anon: a waveform-level front-end (Anon-W2W) and a feature-domain replacement (Anon-W2F). Anon-W2F raises EER by over 3.5x relative to the discrete encoder baseline (11.2% to 41.0%), approaching the 50% random-chance ceiling, while Anon-W2W retains 78-93% of baseline sBERT across setups with sub-second response latency (FRL under 0.8 s).

Vision-language-action (VLA) models rely on discrete action tokenizers to bridge continuous robot control and autoregressive sequence modeling, yet existing tokenizers often trade off between compression, latency, and downstream performance. We revisit this design through the lens of neural audio codecs-convolutional encoder-decoder architectures with residual vector quantization that serve as the standard front end for audio foundation models. Motivated by their success, we introduce the Neural Action Codec (NAC), which treats short robot action trajectories as multi-channel 1D signals and compresses them using a multi-scale RVQGAN architecture. We observe that audio-specific mel-spectrogram objectives are ill-suited for kinematic signals; however, by replacing them with simple time-domain and non-mel spectral reconstruction losses, audio-codec-style models can autoencode actions with high fidelity without substantial architectural changes. NAC provides a compact, ordered token space via offset codebooks, enabling standard autoregressive policies to operate over short, structured sequences. Meanwhile, a Vocos-style decoder with an ISTFT head and adversarial discriminators recovers smooth, detailed trajectories. Across LIBERO-10, RoboMimic, and a suite of real-world manipulation tasks, NAC achieves lower reconstruction error and higher success rates than binning, FAST, and prior VQ-based tokenizers at comparable or better compression rates. These results demonstrate that repurposed neural audio codecs offer a strong, practical backbone for learned action tokenization in modern VLAs.

An avatar that holds a conversation should decide what to say and to move while saying it, yet these abilities live in separate model families: spoken dialogue models produce speech without motion, and co-speech motion models produce motion only from audio handed to them. The standard remedy is a cascade that first generates the spoken response and then runs a motion model over the finished audio, which requires a second full inference pass and precludes any joint optimisation between the two. We present Motion-Omni, an end-to-end framework in which a spoken dialogue model natively outputs explicit facial expression together with hand, upper-body and lower-body motion, generated directly from the hidden states that produce the speech. Joint training is not optional here: with the speech pathway frozen, motion remains misaligned with the audio, and co-adapting the LLM, Speech Generator and Motion Generator under both objectives is what recovers alignment while retaining spoken-dialogue ability. Supervision comes from a scalable, model-agnostic pipeline that pseudo-labels consistent-voice speech responses with a replaceable motion teacher, yielding 422,856 quality-ranked pairs (1,402 hours). We further release SwDA-500 and, to our knowledge, the first public evaluation protocol for stochastic open-ended full-body spoken dialogue, matching audio across motion systems while unifying rendering, automatic metrics, human evaluation, and latency measurement. Instantiated with a Qwen2.5-7B-Instruct backbone, Motion-Omni-Q7 matches the same-audio teacher cascade to within 2% on reference-free motion metrics while responding 5.4 x faster (RTF=0.78, faster than real time), surpasses all non-teacher cascades on beat correlation and diversity, and reaches a 2.62% word error rate, the lowest among the omni-modal systems compared.

This study focuses on a challenging yet promising task, Text-to-Sounding-Video (T2SV) generation, which aims to generate a video with synchronized audio from text conditions, meanwhile ensuring both modalities are aligned with text. Despite progress in joint audio-video training, two critical challenges still remain unaddressed: (1) a single, shared text caption where the text for video is equal to the text for audio often creates modal interference, confusing the pretrained backbones, and (2) the optimal mechanism for cross-modal feature interaction remains unclear. To address these challenges, we first propose the Hierarchical Visual-Grounded Captioning (HVGC) framework that generates pairs of disentangled captions, a video caption, and an audio caption, eliminating interference at the conditioning stage. Based on HVGC, we further introduce BridgeDiT, a novel dual-tower diffusion transformer, which employs a Dual CrossAttention (DCA) mechanism that acts as a robust ``bridge" to enable a symmetric, bidirectional exchange of information, achieving both semantic and temporal synchronization. Extensive experiments on three benchmark datasets, supported by human evaluations, demonstrate that our method achieves state-of-the-art results on most metrics. Comprehensive ablation studies further validate the effectiveness of our contributions, offering key insights for the future T2SV task. All the codes and checkpoints will be publicly released.

Recent diffusion and autoregressive models have substantially improved text-to-video generation, yet producing coherent long-form story videos with consistent characters and aligned audio remains challenging. Existing methods often suffer from narrative drift, unstable character identity, weak cross-scene continuity, and audio-visual mismatch over extended sequences. We propose DramaAgent, a hierarchical, agentic, and model-agnostic framework for long-form text-to-video-and-audio generation. Rather than improving the underlying video backbone itself, DramaAgent introduces an upper-level control layer that decomposes generation into story planning, persistent character conditioning, scene-wise synthesis, and reflection-guided targeted repair. The framework maintains reusable story and character states across scenes, diagnoses failures such as identity drift, missing scene semantics, temporal discontinuity, and cross-modal mismatch, and repairs problematic clips in a stage-specific manner. Experiments across multiple video generation backbones show that DramaAgent improves long-horizon coherence, character consistency, narrative fidelity, and scene-level audio-visual consistency over direct generation and strong baselines. These results suggest that hierarchical agentic control is a practical direction for controllable long-form audiovisual generation. Code: https://github.com/AIGeeksGroup/DramaAgent. Website: https://aigeeksgroup.github.io/DramaAgent.

We introduce Moshi, a speech-text foundation model and full-duplex spoken dialogue framework. Current systems for spoken dialogue rely on pipelines of independent components, namely voice activity detection, speech recognition, textual dialogue and text-to-speech. Such frameworks cannot emulate the experience of real conversations. First, their complexity induces a latency of several seconds between interactions. Second, text being the intermediate modality for dialogue, non-linguistic information that modifies meaning -- such as emotion or non-speech sounds -- is lost in the interaction. Finally, they rely on a segmentation into speaker turns, which does not take into account overlapping speech, interruptions and interjections. Moshi solves these independent issues altogether by casting spoken dialogue as speech-to-speech generation. Starting from a text language model backbone, Moshi generates speech as tokens from the residual quantizer of a neural audio codec, while modeling separately its own speech and that of the user into parallel streams. This allows for the removal of explicit speaker turns, and the modeling of arbitrary conversational dynamics. We moreover extend the hierarchical semantic-to-acoustic token generation of previous work to first predict time-aligned text tokens as a prefix to audio tokens. Not only this "Inner Monologue" method significantly improves the linguistic quality of generated speech, but we also illustrate how it can provide streaming speech recognition and text-to-speech. Our resulting model is the first real-time full-duplex spoken large language model, with a theoretical latency of 160ms, 200ms in practice, and is available at https://github.com/kyutai-labs/moshi.

Existing methodologies for animating portrait images face significant challenges, particularly in handling non-frontal perspectives, rendering dynamic objects around the portrait, and generating immersive, realistic backgrounds. In this paper, we introduce the first application of a pretrained transformer-based video generative model that demonstrates strong generalization capabilities and generates highly dynamic, realistic videos for portrait animation, effectively addressing these challenges. The adoption of a new video backbone model makes previous U-Net-based methods for identity maintenance, audio conditioning, and video extrapolation inapplicable. To address this limitation, we design an identity reference network consisting of a causal 3D VAE combined with a stacked series of transformer layers, ensuring consistent facial identity across video sequences. Additionally, we investigate various speech audio conditioning and motion frame mechanisms to enable the generation of continuous video driven by speech audio. Our method is validated through experiments on benchmark and newly proposed wild datasets, demonstrating substantial improvements over prior methods in generating realistic portraits characterized by diverse orientations within dynamic and immersive scenes. Further visualizations and the source code are available at: https://fudan-generative-vision.github.io/hallo3/.

Diffusion models have emerged as the backbone of modern generative AI, powering advances in vision, language, audio and other modalities. Despite their success, they suffer from hallucinations, implausible samples that lie outside the support of true data distribution, which degrade reliability and trust. In this work, we first empirically confirm previously proposed hypothesis that score smoothness causes hallucinations in Image Generation diffusion models and provide a density-based perspective. We further formalize this notion by linking the hallucinations probability mass to lipschitz constant of the learned score function. Motivated by this, we introduce a Variance-Guided Score Modulation (VSM) strategy that controls the score Jacobian, in turn reducing score smoothness and better approximating the ground truth score that decreases hallucinations. Empirical results on synthetic and real-world datasets demonstrate that our approach reduces hallucinations (up to ~25%) while maintaining high fidelity and diversity, providing a principled step toward more reliable diffusion-based image generation. We also propose two benchmark datasets with extreme semantic variation for systematic hallucination evaluation. Code and Datasets are publicly available at https://github.com/bhosalems/VSM.

Streaming omni-modal models must decide what and when to answer from the video chunks and synchronized audio observed so far. Visual cues often support an interpretation before an utterance or sound event is complete. If that interpretation enters memory as a fact, later reasoning can keep relaying it even after audio contradicts it. We call this failure premature cross-modal commitment. We propose Omni-Streaming Thinking (OST), which generates structured outputs that include evidence observed so far, forecasts of future evidence, and claims based on this evidence. Each claim is initially marked as pending and linked to a future verification interval. Audio and visual evidence are stored separately, and OST checks a claim against the evidence from the specified modality at the end of the verification interval. When contradictory evidence is detected, a refutation process reduces the influence of the claim and its dependent states, and then guides a state update using the new evidence. An answer gate decides whether the answer-critical claims meet the conditions for giving a response. Using a frozen Qwen3-Omni-30B-A3B-Instruct backbone with lightweight adaptation, OST outperforms the strongest open baselines on five streaming and audio-visual benchmarks by more than 10% relative on average. We also introduce OST-DiagBench, which holds video fixed and edits audio to test agreement, absence, contradiction, coexistence, and subtitle-speech conflict. OST reaches d-prime = 2.95, compared with at most 1.38 for open baselines, while reducing vision-induced auditory hallucinations.

Audio-visual understanding remains challenging because models must jointly interpret spoken content, visual events, and their temporal relationships. Existing omni models typically introduce dedicated audio encoders and rely on expensive audio-video-text training, tightly coupling omni capability to specific VLM backbones and potentially weakening their existing visual and reasoning abilities. This raises three questions: whether native omni training is necessary for every new VLM, whether speech-centric omni capability can be added while preserving the original backbone, and where richer acoustic representations remain essential. We introduce Training-Free Omni (TFO), a plug-and-play framework that converts a frozen VLM into a speech-centric omni model without architectural modification, or multimodal re-alignment. TFO uses Whisper to extract confidence-filtered, timestamped transcripts and routes them through the VLM's existing language interface, while leaving its visual pathway unchanged. Across matched comparisons with native omni models on 56 benchmarks and 21 languages, TFO is competitive on audio-visual understanding, improves average audio-only performance across all five model settings, and achieves substantial multilingual speech gains. Freezing the VLM also generally preserves stronger image/video understanding, visual grounding, coding, mathematical reasoning, and medical question answering than the corresponding native omni checkpoints. These results show that strong speech-centric omni understanding can often be obtained through modular audio-to-language routing rather than costly backbone-specific training.

Brain encoding models aim to predict neural responses to stimuli. The most widely used approach fits readouts from frozen pretrained backbones. Using neural data to train large models from scratch or fully fine-tune them after task-pretraining has proven difficult because neural data is costly and limited. We show that parameter-efficient fine-tuning (PEFT) enables us to effectively adapt very large pretrained backbones for brain encoding with a small number of trainable parameters. To make it computationally practical, we exploit the backbone's causal structure to predict multiple timepoints in a single pass instead of using sliding windows, significantly reducing feature-extraction cost. This enables efficient fine-tuning of multimodal backbones with up to 104B parameters, the largest model adapted for brain encoding to our knowledge. Trained with a low-rank adapter (LoRA) and a lightweight readout, our approach consistently improves brain encoding performance across nine diverse multimodal backbones. On the same Qwen3-Omni backbone as the prior state-of-the-art (SoTA), our method BrainLoRA achieves the best published performance on the Algonauts 2025 Challenge. We also show the adapted multimodal backbones remain interpretable: modality attribution recovers the canonical sensory topography that prior work often obtains by construction from separate unimodal encoders, and neural adaptation increases the contribution of audio inputs and cross-modal mixing within the backbone. To scale out the number of adapted models, we develop a multi-LoRA kernel that accelerates joint training of multiple LoRAs on a shared backbone. This enables an ensemble of 200+ adapted models that ranks first on the public Algonauts post-challenge leaderboard at the time of writing. Together, our results highlight PEFT as an efficient and powerful paradigm for brain encoding.

Sound can convey significant information for spatial reasoning in our daily lives. To endow deep networks with such ability, we address the challenge of dense indoor prediction with sound in both 2D and 3D via cross-modal knowledge distillation. In this work, we propose a Spatial Alignment via Matching (SAM) distillation framework that elicits local correspondence between the two modalities in vision-to-audio knowledge transfer. SAM integrates audio features with visually coherent learnable spatial embeddings to resolve inconsistencies in multiple layers of a student model. Our approach does not rely on a specific input representation, allowing for flexibility in the input shapes or dimensions without performance degradation. With a newly curated benchmark named Dense Auditory Prediction of Surroundings (DAPS), we are the first to tackle dense indoor prediction of omnidirectional surroundings in both 2D and 3D with audio observations. Specifically, for audio-based depth estimation, semantic segmentation, and challenging 3D scene reconstruction, the proposed distillation framework consistently achieves state-of-the-art performance across various metrics and backbone architectures.

Recent dense audio-visual (AV) models achieve impressive retrieval and emergent localization, but almost all evidence comes from English-centric, caption-rich web video. It is unclear whether these objectives survive in low-resource, code-switched, and noisy multilingual settings that typify developing regions. We show they do**-**and that the choice of aggregation function becomes even more critical. Using a multilingual subset of Project Vaani spanning dozens of Indian languages and dialectal variants, we compare three contrastive objectives: (i) a global mean-pooled loss (CLIP-style), (ii) a dense max-mean token matcher (DenseAV-style), and (iii) a simple hybrid (motivated by frozen-vision alignment strategies). The dense objective delivers a +59% relative R@1 (Audio Visual) improvement over global pooling and substantially lower mean/median ranks, while consistently producing sharp zero-shot localization heatmaps of spoken objects-despite keeping the vision backbone entirely frozen (no LoRA / partial fine-tuning). Our results demonstrate that dense token routing is not a luxury of high-resource English corpora; it is more decisive when annotations and acoustic cleanliness are scarce. We release the codebase and trained models.

Recent advancements in Multimodal Large Language Models (MLLMs) pursue omni-perception capabilities, yet integrating robust sensory grounding with complex reasoning remains a challenge, particularly for underrepresented regions. In this report, we introduce the research preview of MERaLiON2-Omni (Alpha), a 10B-parameter multilingual omni-perception tailored for Southeast Asia (SEA). We present a progressive training pipeline that explicitly decouples and then integrates "System 1" (Perception) and "System 2" (Reasoning) capabilities. First, we establish a robust Perception Backbone by aligning region-specific audio-visual cues (e.g., Singlish code-switching, local cultural landmarks) with a multilingual LLM through orthogonal modality adaptation. Second, to inject cognitive capabilities without large-scale supervision, we propose a cost-effective Generate-Judge-Refine pipeline. By utilizing a Super-LLM to filter hallucinations and resolve conflicts via a consensus mechanism, we synthesize high-quality silver data that transfers textual Chain-of-Thought reasoning to multimodal scenarios. Comprehensive evaluation on our newly introduced SEA-Omni Benchmark Suite reveals an Efficiency-Stability Paradox: while reasoning acts as a non-linear amplifier for abstract tasks (boosting mathematical and instruction-following performance significantly), it introduces instability in low-level sensory processing. Specifically, we identify Temporal Drift in long-context audio, where extended reasoning desynchronizes the model from acoustic timestamps, and Visual Over-interpretation, where logic overrides pixel-level reality. This report details the architecture, the data-efficient training recipe, and a diagnostic analysis of the trade-offs between robust perception and structured reasoning.

Many existing audio processing and generation models rely on task-specific architectures, resulting in fragmented development efforts and limited extensibility. It is therefore promising to design a unified framework capable of handling multiple tasks, while providing robust instruction and audio understanding and high-quality audio generation. This requires a compatible paradigm design, a powerful backbone, and a high-fidelity audio reconstruction module. To meet these requirements, this technical report introduces QuarkAudio, a decoder-only autoregressive (AR) LM-based generative framework that unifies multiple tasks. The framework includes a unified discrete audio tokenizer, H-Codec, which incorporates self-supervised learning (SSL) representations into the tokenization and reconstruction process. We further propose several improvements to H-Codec, such as a dynamic frame-rate mechanism and extending the audio sampling rate to 48 kHz. QuarkAudio unifies tasks by using task-specific conditional information as the conditioning sequence of the decoder-only LM, and predicting discrete target audio tokens in an AR manner. The framework supports a wide range of audio processing and generation tasks, including speech restoration (SR), target speaker extraction (TSE), speech separation (SS), voice conversion (VC), and language-queried audio source separation (LASS). In addition, we extend downstream tasks to universal free-form audio editing guided by natural language instructions (including speech semantic editing and audio event editing). Experimental results show that H-Codec achieves high-quality audio reconstruction with a low frame rate, improving both the efficiency and performance of downstream audio generation, and that QuarkAudio delivers competitive or comparable performance to state-of-the-art task-specific or multi-task systems across multiple tasks.

Recent advances in spoken dialogue language models have shifted from turn-based to full-duplex designs, where the model continuously listens to the user while generating responses. However, existing duplex backbones still lack a native channel for in-conversation planning and tool calling, leaving real-time agentic behaviour either tied to turn boundaries or relegated to an external cascade. We propose DuplexSLA, a native full-duplex Speech-Language-Action foundation model that decodes assistant audio together with a structured action stream on a shared 160 ms chunk timeline. DuplexSLA is built on a dual-stream three-channel formulation: a continuous user audio channel, a discrete assistant audio channel, and a rate-limited textual action channel, all decoded jointly by a single backbone, so that listening, speaking, planning, and tool calling unfold on one shared clock. Two capabilities define the model: (1) semantic-driven turn-taking control, where interruption, pause, and backchannel are handled inside the same backbone instead of by an external semantic VAD; and (2) in-conversation planning and tool calling, where planning text and structured tool calls are emitted on the action channel without halting assistant audio, so that multi-action and backchannel-triggered tool use are interleaved with ongoing speech. To evaluate these capabilities together, we further construct DuplexSLA-Bench, a duplex benchmark covering pause, interrupt, and backchannel turn-taking together with three styles of in-conversation tool calling. Our project page, interactive demos, and the DuplexSLA-Bench evaluation suite are publicly available at https://github.com/hyzhang24/DuplexSLA.

We describe our entry to the efficiency track of the Academic Text-to-Music (ATTM) Grand Challenge at ICME 2026. Beyond the challenge protocol's FAD-CLAP and CLAP score, we add a learned human-preference reward from TuneJury, a twin pairwise ranker trained over open music-preference datasets. The reward serves both as a training-time conditioning signal and as a sample-selection criterion. The pipeline combines five engineering decisions on a 120M-parameter FluxAudio-S backbone, four at training time and one at inference: (i) training-time reward conditioning that doubles as an inference-time CFG axis, (ii) a sweep over five score-conditioning architectures, where training and inference use different variants, (iii) expert iteration on the top decile, (iv) a short preference-tuning pass (CRPO) for audio-text alignment, and (v) inference post-processing via joint CFG, source separation, and loudness normalization. Per-stage decomposition on 100 Song Describer prompts shows training-time reward conditioning as a functional conditioning axis, expert iteration as the dominant contributor, the preference-tuning pass adding only noise-level gain, and the inference-time score scalar already saturated by the end of the chain.

Speech emotion recognition (SER) is commonly formulated as utterance-level classification, although conversational emotion depends on a speaker's usual vocal range and the emotional context established by previous utterances. Speech-language models provide strong pretrained acoustic and semantic representations, and can adapts them to SER labels via finetune, but this mechanism still missing per-dialogue state. We study whether test-time neural memory can supply this missing context while leaving the large audio language models (LALMs) backbone intact. Building on Titans, we introduce a plug-and-play Memory-as-a-Layer (MAL) adapter that writes dialogue history into a small neural memory and reads it back as an audio-token-aligned residual update, avoiding changes to the host model's token positions. Across different audio LLMs and emotion recognition datasets evaluations, our design improves SER performs across different evaluation metrics, supporting test-time memory as a residual contextual mechanism for conversational SER.

Recent neural audio codec-based speech generation (CodecFake) produces highly realistic audio, posing a challenge to existing deepfake countermeasure models. While using codec resynthesized speech (CoRS) as proxy data improves performance, it often suffers from limited generalization. We propose Domain-Shift Feature Augmentation (DSFA), which simulates "in-the-wild" variations by transforming deterministic feature statistics into stochastic distributions during fine-tuning. To evaluate generalization, we further introduce Codec-based Speech Generation Extension Evaluation (CoSG ExtEval) dataset, a more challenging extension of the CoSG Eval (from CodecFake+) dataset, featuring 40 unseen generative models and long-form audio. Experimental results demonstrate that combining a post-trained SSL backbone with DSFA effectively narrows the proxy-to-wild domain gap. This approach achieves state-of-the-art performance across diverse CodecFake attacks in both CoSG Eval and CoSG ExtEval.

Most successful self-supervised learning methods are trained to align the representations of two independent views from the data. State-of-the-art methods in video are inspired by image techniques, where these two views are similarly extracted by cropping and augmenting the resulting crop. However, these methods miss a crucial element in the video domain: time. We introduce BraVe, a self-supervised learning framework for video. In BraVe, one of the views has access to a narrow temporal window of the video while the other view has a broad access to the video content. Our models learn to generalise from the narrow view to the general content of the video. Furthermore, BraVe processes the views with different backbones, enabling the use of alternative augmentations or modalities into the broad view such as optical flow, randomly convolved RGB frames, audio or their combinations. We demonstrate that BraVe achieves state-of-the-art results in self-supervised representation learning on standard video and audio classification benchmarks including UCF101, HMDB51, Kinetics, ESC-50 and AudioSet.

Gestures are non-verbal but important behaviors accompanying people's speech. While previous methods are able to generate speech rhythm-synchronized gestures, the semantic context of the speech is generally lacking in the gesticulations. Although semantic gestures do not occur very regularly in human speech, they are indeed the key for the audience to understand the speech context in a more immersive environment. Hence, we introduce LivelySpeaker, a framework that realizes semantics-aware co-speech gesture generation and offers several control handles. In particular, our method decouples the task into two stages: script-based gesture generation and audio-guided rhythm refinement. Specifically, the script-based gesture generation leverages the pre-trained CLIP text embeddings as the guidance for generating gestures that are highly semantically aligned with the script. Then, we devise a simple but effective diffusion-based gesture generation backbone simply using pure MLPs, that is conditioned on only audio signals and learns to gesticulate with realistic motions. We utilize such powerful prior to rhyme the script-guided gestures with the audio signals, notably in a zero-shot setting. Our novel two-stage generation framework also enables several applications, such as changing the gesticulation style, editing the co-speech gestures via textual prompting, and controlling the semantic awareness and rhythm alignment with guided diffusion. Extensive experiments demonstrate the advantages of the proposed framework over competing methods. In addition, our core diffusion-based generative model also achieves state-of-the-art performance on two benchmarks. The code and model will be released to facilitate future research.

We present XKD, a novel self-supervised framework to learn meaningful representations from unlabelled videos. XKD is trained with two pseudo objectives. First, masked data reconstruction is performed to learn modality-specific representations from audio and visual streams. Next, self-supervised cross-modal knowledge distillation is performed between the two modalities through a teacher-student setup to learn complementary information. We introduce a novel domain alignment strategy to tackle domain discrepancy between audio and visual modalities enabling effective cross-modal knowledge distillation. Additionally, to develop a general-purpose network capable of handling both audio and visual streams, modality-agnostic variants of XKD are introduced, which use the same pretrained backbone for different audio and visual tasks. Our proposed cross-modal knowledge distillation improves video action classification by 8% to 14% on UCF101, HMDB51, and Kinetics400. Additionally, XKD improves multimodal action classification by 5.5% on Kinetics-Sound. XKD shows state-of-the-art performance in sound classification on ESC50, achieving top-1 accuracy of 96.5%.