You're looking at a specific version of this model. Jump to the model overview.
Input schema
The fields you can use to run this model with an API. If you don’t give a value for a field its default value will be used.
| Field | Type | Default value | Description |
|---|---|---|---|
| audio |
string
|
Upload the file to be processed here.
|
|
| model |
None
|
htdemucs
|
Choose the demucs audio that proccesses your audio. Options: htdemucs (first version of hybrid transformer demucs), htdemucs_ft (fine-tuned version of htdemucs, separation will take 4 times longer but may be a bit better), htdemucs_6s (adds piano and guitar sources to htdemucs), hdemucs_mmi (hybrid demucs v3), mdx (trained on exclusively MusDB HQ), mdx_q (quantized version of mdx, slightly faster but worse quality), mdx_extra (adds extra training data to mdx, this is what the cog previously used by default), mdx_extra_q (quantized version of mdx_extra, slightly faster but worse quality)
|
| two_stems |
None
|
If you just want to isolate one stem, you can choose it here. This does not improve performance, as it just combines all of the stems that you did not choose.
|
|
| output_format |
None
|
mp3
|
Choose the audio format you would like the result to be returned in.
|
| split |
boolean
|
True
|
Choose whether or not the audio should be split into chunks.
|
| segment |
integer
|
Choose the size of each chunk. This only has an effect if split is set to true.
|
|
| clip_mode |
None
|
rescale
|
Choose the strategy for avoiding clipping. Rescaling adjusts the overall scale of a signal to prevent any clipping, while hard clipping limits the signal to a maximum range, distorting parts of the signal that exceed that range.
|
| shifts |
integer
|
1
|
Choose the amount random shifts for equivariant stabilization. This performs multiple predictions with random shifts of the input and averages them, which makes it x times slower.
|
| overlap |
number
|
0.25
|
Choose the amount of overlap between prediction windows.
|
Output schema
The shape of the response you’ll get when you run this model with an API.
Schema
{'properties': {'stems': {'items': {'properties': {'audio': {'format': 'uri',
'title': 'Audio',
'type': 'string'},
'name': {'title': 'Name',
'type': 'string'}},
'required': ['name', 'audio'],
'title': 'DemucsStem',
'type': 'object'},
'title': 'Stems',
'type': 'array'}},
'required': ['stems'],
'title': 'DemucsResponse',
'type': 'object'}