Instrument Separation

Split a song into its individual components — vocals, drums, bass, guitar, keys, and more. Use the separated stems for remixing, rebalancing, sampling, or feeding into downstream production workflows. See the full list of available stem models on the Models page.

Create a Task

Include one target per stem you want to extract:

import requests

API_KEY = "your_api_key"
HEADERS = {"Content-Type": "application/json", "x-api-key": API_KEY}

response = requests.post(
    "https://api.audioshake.ai/tasks",
    headers=HEADERS,
    json={
        "assetId": "your_asset_id",
        "targets": [
            {"model": "vocals", "formats": ["wav"]},
            {"model": "drums", "formats": ["wav"]},
            {"model": "bass", "formats": ["wav"]},
            {"model": "other", "formats": ["wav"]}
        ]
    }
)

task_id = response.json()["id"]
print(f"Task created: {task_id}")

const API_KEY = "your_api_key";
const headers = { "Content-Type": "application/json", "x-api-key": API_KEY };

const createRes = await fetch("https://api.audioshake.ai/tasks", {
  method: "POST",
  headers,
  body: JSON.stringify({
    assetId: "your_asset_id",
    targets: [
      { model: "vocals", formats: ["wav"] },
      { model: "drums", formats: ["wav"] },
      { model: "bass", formats: ["wav"] },
      { model: "other", formats: ["wav"] }
    ]
  })
});

const { id: taskId } = await createRes.json();
console.log(`Task created: ${taskId}`);

curl -X POST "https://api.audioshake.ai/tasks" \
  -H "Content-Type: application/json" \
  -H "x-api-key: $AUDIOSHAKE_API_KEY" \
  -d '{
    "assetId": "your_asset_id",
    "targets": [
      { "model": "vocals", "formats": ["wav"] },
      { "model": "drums", "formats": ["wav"] },
      { "model": "bass", "formats": ["wav"] },
      { "model": "other", "formats": ["wav"] }
    ]
  }'

Check Task status to monitor progress and download results, or use webhooks to be notified when each target completes.

Get everything else

Add a residual target to receive one stem containing everything the task’s other targets don’t cover:

"targets": [
  { "model": "guitar", "formats": ["wav"] },
  { "model": "bass", "formats": ["wav"] },
  { "model": "drums", "formats": ["wav"] },
  { "model": "residual", "formats": ["wav"] }
]

All targets must use the same formats, and at least one non-residual target is required.

This is different from the per-target residual: true option, which returns everything outside that one stem. A task can use one or the other, not both.

Drum kit separation

The drum_kit model breaks the drums into six stems from a single target: kick, snare, toms, hihat, cymbals, and percussions.

{ "model": "drum_kit", "formats": ["wav"] }

Use cases

Build instrumentals by removing the vocal stem
Rebalance drum and bass levels for a custom mix
Export individual stems into a DAW for editing
Create sample packs from existing recordings

Build Karaoke Tracks

Combine instrumental + lyrics for karaoke publishing.

Dialogue Separation

Separate speech from music and effects for post-production.

Getting started

Tutorials

Tools

Support

Create a Task

Get everything else

Drum kit separation

Use cases

Build Karaoke Tracks

Dialogue Separation

​Create a Task

​Get everything else

​Drum kit separation

​Use cases

Build Karaoke Tracks

Dialogue Separation

Create a Task

Get everything else

Drum kit separation

Use cases