Add/audio stt #507
Add/audio stt #507
Conversation
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
Codecov Report❌ Patch coverage is
📢 Thoughts on this report? Let us know! |
|
|
||
|
|
||
| class AudioRequest(BaseModel): | ||
| class AudioSpeechRequest(BaseModel): |
There was a problem hiding this comment.
Preference: Maybe we call this AudioGenerationRequest just to make it clear what it is? Good call renaming as part of this change.
| "version": "0.0.7", | ||
| "supports": ["Text-to-Speech", "Audio"], | ||
| "model_architectures": ["MLXTextToSpeech", "StyleTTS2"], | ||
| "version": "0.0.8", |
There was a problem hiding this comment.
This is a big change. I'd make it at least 0.1.0. :D
| stream = params.get("stream", False) | ||
|
|
||
| experiment_dir = get_experiments_dir() | ||
| audio_dir = params.get("audio_dir", None) |
There was a problem hiding this comment.
There's a security alert around this audio_dir stuff which I think is only part of this PR because of an indentation change.
It's probably not a bad idea to implement this change though. I think the AI suggested code is fine.
dadmobile
left a comment
dadmobile
left a comment
There was a problem hiding this comment.
I couldn't start the whisper-tiny-mlx model because it didn't read in the architecture. So it thought this should run on mlx_server instead of mlx_audio_server? How do we fix this for other models...add to our gallery and set the architecture?
|
@mina Parham to implement changes, then add model + APP |
|
Wrong mina.
|
… in path expression Co-authored-by: Copilot Autofix powered by AI <62310815+github-advanced-security[bot]@users.noreply.github.com>
the app branch to test this is: transformerlab/transformerlab-app#719
the model to use:
mlx-community/whisper-tiny-mlx