Spec for audio to audio #502
Conversation
Wauplin
requested review from
osanseviero,
SBrandeis and
gary149
as code owners
|
Should generate TS type with the helper command ( |
Done in f80bf58. |
There was a problem hiding this comment.
Is the pipeline generating several audios at once?
If not I would recommend not wrapping the output in an array for consistency with other tasks (see ImageToText for example)
https://github.com/huggingface/huggingface.js/blob/af447753043472d8ba65ab0591be7487433f7261/packages/tasks/src/tasks/image-to-text/spec/output.json#L1-L15
There was a problem hiding this comment.
Yes it does. You can try on the widget in https://huggingface.co/tasks/audio-to-audio.
| "type": "string", | ||
| "description": "The label of the audio file." | ||
| }, | ||
| "content-type": { |
There was a problem hiding this comment.
Maybe?
Can you link to where those parameters are documented please?
| "content-type": { | |
| "content_type": { |
There was a problem hiding this comment.
It's not documented AFAIK. I only know it from the call made by the widget on https://huggingface.co/tasks/audio-to-audio 😕
There was a problem hiding this comment.
For content-type vs content_type I've added a normalizer rule in Python to accept both since the Python attribute can only be content_type (and is generated like this).
There was a problem hiding this comment.
You can see some specification from https://github.com/huggingface/api-inference-community/blob/main/docker_images/speechbrain/app/pipelines/audio_to_audio.py and potential outputs
| * | ||
| * A generated audio file with its label. | ||
| */ | ||
| export interface AudioToAudioOutputElement { |
There was a problem hiding this comment.
FYI here is the returned values from community API https://github.com/huggingface/api-inference-community/blob/main/docker_images/speechbrain/app/pipelines/audio_to_audio.py#L37-L44
Json schema for audio-to-audio task. This is the "expected type" in the Python client but would prefer to double-check it's correct before continuing. Ping @Vaibhavs10 @osanseviero could you have a look please?
(useful for huggingface/huggingface_hub#2036)