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 |
|---|---|---|---|
| song |
string
|
The song to cover (any format; stereo is kept). An a cappella vocal works too.
|
|
| voice |
string
|
The RVC voice: a .zip holding a .pth (and usually an .index), or a .pth.
|
|
| voice_octave |
None
|
auto
|
Octave for the new voice. auto picks -1, 0 or +1 from the voice's trained pitch (or voice_f0_hz) against the singer's.
|
| voice_f0_hz |
number
|
0
Max: 1000 |
The voice's typical pitch in Hz when the voice file does not record it; 0 = unknown.
|
| key_shift |
integer
|
0
Min: -12 Max: 12 |
Semitones to transpose the whole song: vocals, backing vocals and instrumental.
|
| pitch_extractor |
None
|
rmvpe
|
Pitch extraction for the conversion.
|
| index_rate |
number
|
0.5
Max: 1 |
How much the voice's retrieval index shapes the timbre (more = closer to the voice, less = clearer words).
|
| protect |
number
|
0.33
Max: 0.5 |
Protect breaths and voiceless consonants from conversion artefacts (0.5 = off).
|
| rms_mix_rate |
number
|
0.25
Max: 1 |
0 follows the singer's loudness contour, 1 keeps the converted voice's own.
|
| separation |
None
|
auto
|
auto: separate when the input has a backing track. never: the input is already a vocal.
|
| backing_vocals |
None
|
keep
|
Harmonies and backing vocals: keep them (original voices) or drop them.
|
| remove_reverb |
boolean
|
True
|
De-reverb the lead before conversion (strongly recommended).
|
| lead_volume_db |
number
|
0
Min: -20 Max: 20 |
New lead vocal level against the original lead, in dB.
|
| backing_volume_db |
number
|
0
Min: -20 Max: 20 |
Backing vocals level, in dB.
|
| instrumental_volume_db |
number
|
0
Min: -20 Max: 20 |
Instrumental level, in dB.
|
| reverb_size |
number
|
0.15
Max: 1 |
Reverb room size.
|
| reverb_wetness |
number
|
0.2
Max: 1 |
Reverb wet level.
|
| reverb_dryness |
number
|
0.8
Max: 1 |
Reverb dry level.
|
| reverb_damping |
number
|
0.7
Max: 1 |
Reverb high-frequency damping.
|
| output_format |
None
|
mp3
|
mp3 is 320 kbps; wav and flac are 16-bit. All 44.1 kHz stereo.
|
| return_stems |
boolean
|
False
|
Also return the new lead vocal and the instrumental as separate files.
|
Output schema
The shape of the response you’ll get when you run this model with an API.
{'items': {'format': 'uri', 'type': 'string'},
'title': 'Output',
'type': 'array'}