Audio
MOSS-TTSD
Multilingual long-form dialogue speech with zero-shot voice cloning.
★ 1.4k starsApache-2.0
View guide →CosyVoice: multilingual voice generation with zero-shot cloning, streaming and fine-tuning
CosyVoice is an open-source audio project by QwenAudio (QwenAudio/CosyVoice on GitHub). It has 23,892 GitHub stars and 2,723 forks, is released under the Apache-2.0 license, and was last updated on GitHub on 25 May 2026.
Decide what you want to achieve with CosyVoice and check the README on GitHub for requirements before installing.
Clone the repository from GitHub (or take the packaged install route in step 3 if the README offers one).
Follow the project's documented install commands — the Run section below lists the exact ones — and add any required API keys or models.
Work through the README's quickstart example end to end, then adapt it to your own project or data.
Models cover Chinese, English, Japanese, Korean, Cantonese and more, including cross-lingual cloning.
Model variants support zero-shot cloning, supervised fine-tuned voices and instruction-controlled synthesis.
Beyond inference, the repo includes training/fine-tuning code and deployment paths.
Run inference through vLLM for throughput, or serve via gRPC/FastAPI examples.
The CosyVoice line is designed for low-latency, streaming TTS use cases.
A model zoo (including 300M-class models) lets you balance quality against hardware.
Clone with submodules, as the README's install section specifies (note the repo also lives under FunAudioLLM). The README asks for a recursive clone for submodules: run `git submodule update --init --recursive` after cloning.
git clone https://github.com/QwenAudio/CosyVoice.git
cd CosyVoiceInstall the Python dependencies from requirements.txt (the README suggests a mirror for faster downloads in China).
pip install -r requirements.txtDownload a CosyVoice model with the ModelScope or Hugging Face SDK snippet from the README, then run the basic-usage example for SFT, instruct or zero-shot inference.
# see the README 'Model download' and 'Basic Usage' sections for the exact snippetsCosyVoice is maintained by the QwenAudio organization/account on GitHub, together with open-source contributors, in the public repository QwenAudio/CosyVoice.
Yes — the code is open source under the Apache-2.0 license. You can use, modify and redistribute it under that license's terms.
No API key — models download from ModelScope/Hugging Face and run on your GPU; budget for GPU compute for real-time use.
OpenVoice focuses on instant cloning simplicity under MIT; CosyVoice offers a broader multilingual model family with training and production serving paths, from an Alibaba research team.
Clone the repository from GitHub, follow the install commands in the Run section (taken from the project's README), and complete the documented quickstart example before adapting it to your own use case.
Multilingual long-form dialogue speech with zero-shot voice cloning.
Gradio WebUI for creators and developers, featuring key TTS (Edge-TTS, kokoro) and zero-shot Voice Cloning (E2 & F5-TTS, CosyVoice), with Whisper audio proce…
IndexTTS: industrial-level, controllable zero-shot text-to-speech with voice cloning
Voice cloning inside Kokoro: natural multilingual speech, your voice.
Multilingual TTS with voice cloning and duration control, on T5Gemma.
VoxCPM2: tokenizer-free TTS for multilingual speech, creative voice design and true-to-life cloning
The most-starred other Audio projects in the LibreAI catalogue — the realistic shortlist if CosyVoice is not the right fit.
利用 AI 大模型和自动化工作流，根据主题或关键词一键生成高清短视频。Generate HD short videos from a topic or keyword with an automated AI workflow.
Robust Speech Recognition via Large-Scale Weak Supervision
1 min voice data can also be used to train a good TTS model! (few shot voice cloning)
Clone a voice from 5 seconds of audio and generate arbitrary speech in real time
Port of OpenAI's Whisper model in C/C++
Free and open — grab the source, star the repo, and start creating.