ASLP-lab commited on
Commit
0eb78ac
·
verified ·
1 Parent(s): 6ae9258

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +56 -0
README.md CHANGED
@@ -1,3 +1,59 @@
1
  ---
2
  license: apache-2.0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: apache-2.0
3
+ pipeline_tag: text-to-speech
4
+ language:
5
+ - zh
6
+ - en
7
+ - de
8
+ - fr
9
+ - ja
10
+ - ko
11
+ - yue
12
+ tags:
13
+ - m5-tts
14
+ - voice-cloning
15
+ - multilingual
16
+ - pytorch
17
  ---
18
+
19
+ # M5-TTS
20
+
21
+ English | [简体中文](README_zh.md)
22
+
23
+ [GitHub](https://github.com/ASLP-lab/M5-TTS) · [Demo](https://aslp-lab.github.io/M5-TTS-Demo-Page/) · [Space](https://huggingface.co/spaces/ASLP-lab/M5-TTS-Space) · [Models](https://huggingface.co/collections/ASLP-lab/m5-tts)
24
+
25
+ **M5-TTS** is a compact multilingual and multi-dialect text-to-speech system from ASLP@NPU and LycheeAI Lab. It combines a 30M temporal backbone with a 20M local multi-token prediction decoder, using 16 audio codebooks at 12.5 frames/s for **24 kHz** speech generation. The speech codec is separate.
26
+
27
+ This repository provides the reference-based voice-cloning checkpoint. It supports Chinese, English, German, French, Japanese and Korean, with Cantonese and Sichuan dialect conditioning. For preset-speaker synthesis without reference audio, use [M5-TTS-50M-CustomVoice](https://huggingface.co/ASLP-lab/M5-TTS-50M-CustomVoice).
28
+
29
+ ## Quickstart
30
+
31
+ Use Python 3.10 on Linux with a BF16-capable NVIDIA GPU and a compatible CUDA driver. Install from the code repository:
32
+
33
+ ```bash
34
+ git clone https://github.com/ASLP-lab/M5-TTS.git
35
+ cd M5-TTS
36
+ conda create -n m5tts python=3.10 -y
37
+ conda activate m5tts
38
+ pip install -r requirements.txt
39
+ pip install --no-build-isolation flash-attn==2.7.3
40
+ ```
41
+
42
+ Run from the repository root:
43
+
44
+ ```bash
45
+ python infer_clone.py \
46
+ --model ASLP-lab/M5-TTS \
47
+ --text "你好,欢迎使用我们的语音合成模型。" \
48
+ --prompt-wav assets/aslp_default_voice.wav \
49
+ --prompt-text "猪携带该疾病,然后通过蚊子传染给人类。" \
50
+ --output outputs/clone.wav
51
+ ```
52
+
53
+ Replace the reference recording and its transcript together to clone another voice. The M5 checkpoint, text tokenizer and [audio tokenizer](https://huggingface.co/ASLP-lab/M5-TTS-Audio-Tokenizer) are downloaded automatically. Use `--lang` for the target language; Cantonese uses `--lang Chinese --accent yue`, and Sichuan uses `--lang Chinese --accent chuan`.
54
+
55
+ For full usage and evaluation results, see the [project README](https://github.com/ASLP-lab/M5-TTS#quickstart). Only clone voices you own or have permission to use.
56
+
57
+ ## License
58
+
59
+ Apache-2.0. The Qwen-derived audio codec retains its upstream attribution; see the [third-party notice](https://github.com/ASLP-lab/M5-TTS/blob/main/m5/codec_qwen12hz/NOTICE).