Qwen/Qwen3-TTS-12Hz-1.7B-Base

2B • Updated Jan 23 • 3.19M • 553

Note Qwen3-TTS base checkpoint for reference-audio voice cloning and adaptation. Multilingual speech generation. Apache-2.0.

hexgrad/Kokoro-82M

Text-to-Speech • Updated Apr 10, 2025 • 10.6M • • 7.24k

Note Lightweight 82M-parameter TTS baseline with a broad deployment ecosystem. Useful when model size matters. Apache-2.0.

ResembleAI/chatterbox-turbo

Text-to-Speech • Updated Dec 15, 2025 • • 697

Note English TTS with voice cloning; the Turbo variant focuses on efficient generation. MIT.

kyutai/tts-1.6b-en_fr

Text-to-Speech • Updated Sep 11, 2025 • 159k • 384

Note Streaming English/French TTS, including dialogs. Voice conditioning through precomputed voice embeddings. CC-BY-4.0.

fishaudio/s2-pro

Text-to-Speech • 5B • Updated Mar 11 • 33.4k • 1.42k

Note Multilingual TTS with inline prosody/emotion control, multi-speaker generation and streaming inference. Fish Audio Research License; check terms for intended use.

FunAudioLLM/Fun-CosyVoice3-0.5B-2512

Text-to-Speech • Updated Feb 3 • 186k • 660

Note Zero-shot cross-lingual voice cloning across 9 languages plus Chinese dialects. Streaming text input/audio output and pronunciation control. Apache-2.0.

SWivid/F5-TTS

Text-to-Speech • Updated Mar 21, 2025 • 823k • 1.22k

Note Flow-matching TTS research baseline; repository includes F5TTS_v1_Base and earlier checkpoints. Pretrained weights are CC-BY-NC-4.0 (noncommercial).

nari-labs/Dia-1.6B-0626

Text-to-Speech • 2B • Updated Jul 3, 2025 • 10.8k • 133

Note Transcript-to-dialogue synthesis with speaker tags, audio conditioning, voice cloning and nonverbal sounds. Scripted dialogue generation. Apache-2.0.

microsoft/VibeVoice-Realtime-0.5B

Text-to-Speech • 1B • Updated Dec 12, 2025 • 255k • 1.3k

Note 0.5B TTS with streaming text input and long-form audio generation. Single speaker; intended for English, other languages are exploratory. MIT metadata; see card usage terms.

parler-tts/parler-tts-mini-v1

Text-to-Speech • 0.9B • Updated Nov 25, 2024 • 14.9k • 153

Note English TTS controlled with natural-language descriptions of voice, pace, pitch and recording conditions. Useful prompt-controlled baseline. Apache-2.0.