gpustack / gpustack/gpustack

CosyVoice Model

Open
#5,000 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
5.7k
Forks
652
Avg merge
1d 8h
Merged PRs (30d)
57

Description

There are a few questions:
When using the CosyVoice sound model, voice cloning is not possible. I provided reference audio and text, but even with identical parameters as the native method, it remains ineffective—the generated voice does not match the original. The native approach works normally, but cloning fails in GPUStack;
2. GGUF cannot be tested with benchmarks, as neither ASR nor TTS can be used for testing;
3. Does not support backend file mistralai/Voxtral-Mini-4B-Realtime-2602

有几个问题:
1、使用声音模型CosyVoice的时候无法克隆声音,我提供了参考声音和参考文本,即使是提供了与原生相同的参数也是无效的,生成的声音不是我提供的声音;使用原生方法是可以正常克隆的,但在GPUStack中无法克隆;
2、gguf不能使用基准测试,ASR和TTS都无法测试;
3、不支持后端文件mistralai/Voxtral-Mini-4B-Realtime-2602

Image Image Image Image Image

There are a few questions:
When using the CosyVoice sound model, voice cloning is not possible. I provided reference audio and text, but even with identical parameters as the native method, it remains ineffective—the generated voice does not match the original. The native approach works normally, but cloning fails in GPUStack;
2. GGUF cannot be tested with benchmarks, as neither ASR nor TTS can be used for testing;
3. Does not support backend file mistralai/Voxtral-Mini-4B-Realtime-2602

There are several questions:

  1. The sound cannot be cloned when using the sound model CosyVoice. I provided a reference sound and reference text. Even if I provide the same parameters as the original one, it is invalid. The generated sound is not the sound I provided. It can be cloned normally using the native method, but it cannot be cloned in GPUStack;
  2. gguf cannot be used for benchmark testing, and neither ASR nor TTS can be tested;
  3. The backend file mistralai/Voxtral-Mini-4B-Realtime-2602 is not supported.
Image Image Image Image Image

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

The report names no repository files, tests, or entry points. Start by separating and reproducing the CosyVoice cloning failure, GGUF benchmark limitation, and Voxtral backend support issue, then trace each through the relevant model-serving and benchmark paths. Done means each reported behavior has a confirmed resolution or a documented supported limitation.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
ai, backend, testing-qa
Issue type
Bug
Difficulty
5/5
Estimated time
Over a week
Activity status
Quiet
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.