Skip to main content
POST
Get Vox Stem
Extract the vocal stem from an audio clip. The returned id is the vox_audio_id for Create Persona — the first step when building a Vox persona for use with task=inspiration and persona_style: "vox".
This endpoint returns the result directly (not a task ID). The extraction is processed synchronously.
By default the service extracts roughly 45–74 seconds of vocals. If the song is shorter than 74s, or you need a different range, coordinate with your integration for custom vocal_start_s / vocal_end_s handling where supported.

Request Body

string
required
The audio clip ID to extract vocals from.

Response

string
The vox audio ID. Use this as vox_audio_id when creating a persona.
string
The extraction status (e.g., "complete").
string
The original clip ID that was processed.
number
Detected vocal start time in seconds.
number
Detected vocal end time in seconds.
string
URL to the extracted vocal audio file.

Example

Response Example

Workflow: Creating a Vox Persona

  1. Get Vox Stem — extract vocals; save returned id as vox_audio_id
  2. Create Persona — pass root_clip_id as clip_id, plus vocal range on the root clip
For persona creation, vocal_start_s / vocal_end_s define the valid vocal range on the root clip (often 0 through the clip’s full length). They are not the same as the internal 45–74s slice used by getVoxStem.
Use the persona in Inspiration generation with persona_style: "vox" and artist_clip_id set to the persona’s root_clip_id.