AvailableAsynchronous source separation

Audio Separation

Turn one mixed track into three useful layers.

Submit audio once, follow a durable task, and retrieve separate dialogue, music, and effects tracks for the next step in your workflow.

Focused input. Predictable output.
InputAudio
Output3 tracks
Source separation

Workflow

A direct path from request to result.

Each product keeps the interface focused on one concrete audio task.

01

Upload the mix

Submit MP3, MP4/M4A, WAV, or WebM audio up to 100 MB.

02

Poll the task

Track the job from queued through in-progress processing.

03

Collect the tracks

Download dialogue, music, and effects from the completed result.

Capabilities

Built around Audio Separation.

Separate dialogue, music and effects into task-ready audio tracks.

Dialogue

Isolate spoken content for transcription, translation, or editing.

Music

Extract the musical layer for replacement, analysis, or reuse.

Effects

Keep environmental and sound-effect layers separate from speech.

API

A contract designed for developers.

Create a task, then poll its ID until it reaches a terminal state.

Task states

queuedin_progresscompletedfailed

Completed output

DialogueMusicEffects
RequestPOST /v1/audio/separations
curl https://taskinfer.com/v1/audio/separations \
+  -H "Authorization: Bearer $TASKINFER_API_KEY" \
+  -F [email protected]

Usage-based pricing

20,000 credits per started minute

Each started audio minute is billed as one whole minute.

View all pricing

Go deeper

Read the full request and response contract.

Open Audio Separation documentation