AudioShake API — Hackathon Examples

Simple Python examples for the AudioShake stem separation API. Upload any song and get back isolated vocals, drums, bass, guitar, lyrics, and more.

Quick Start

1. Get an API Key

Sign up at dashboard.audioshake.ai → Settings → API Keys → Create new key.

2. Set Up

git clone https://github.com/audioshake/hackathon-examples.git
cd hackathon-examples

pip install -r requirements.txt

# Set your API key
cp .env.example .env
# Edit .env and paste your key

3. Run an Example

# Try it with a demo track (no file needed)
python 01_vocals_and_instrumental.py https://demos.audioshake.ai/demo-assets/shakeitup.mp3

# Or use your own file
python 01_vocals_and_instrumental.py my_song.mp3

Examples

#	Script	What it does
01	01_vocals_and_instrumental.py	Split a song into vocals + instrumental. Great for karaoke or remixes.
02	02_full_band_stems.py	Get individual stems: vocals, drums, bass, and other instruments.
03	03_lyrics_transcription.py	Isolate vocals and transcribe lyrics with timestamps (TXT, JSON).

How It Works

All examples use the AudioShakeClient in audioshake_client.py, which wraps the Tasks API:

from audioshake_client import AudioShakeClient

client = AudioShakeClient(api_key="your_key")

task = client.process(
    source_url="https://example.com/song.mp3",
    targets=[
        {"model": "vocals", "formats": ["wav"]},
        {"model": "drums", "formats": ["wav"]},
    ],
    output_dir="output",
)

The workflow is: Upload → Create Task → Poll → Download. The process() method handles all of it.

Available Models

Instrument Stems

Model	Description
vocals	All vocals (lead + backing)
vocals_lead	Lead vocal only
vocals_backing	Backing vocals only
instrumental	Full mix minus vocals
drums	Drums and percussion
bass	Bass
guitar	All guitar
guitar_electric	Electric guitar
guitar_acoustic	Acoustic guitar
piano	Piano
keys	All keyboard instruments
strings	Strings
wind	Woodwinds and brass
other	Everything except vocals, drums, bass

Speech & Post-Production

Model	Description
dialogue	Isolate speech from background noise/music
speech_clarity	Clean up noisy speech recordings
effects	Sound effects with dialogue and music removed
multi_voice	Separate individual speakers

Lyrics & Transcription

Model	Description
transcription	Lyrics/speech with word-level timestamps
alignment	Align timestamps to an existing transcript

Copyright & Detection

Model	Description
music_detection	Detect where music appears (returns timestamps)
music_removal	Remove music, keep speech + effects

Hackathon Ideas

• Stem Mixer Web App — Upload a song, separate stems, add per-stem volume controls
• AI Karaoke — Remove vocals, display synced lyrics from transcription
• Podcast Cleaner — Use dialogue or speech_clarity to clean up noisy recordings
• Sample Finder — Isolate drums/bass/vocals from tracks to build a sample library
• Music Education Tool — Isolate individual instruments for practice or analysis
• Remix Machine — Separate stems, apply effects, recombine

API Reference

Full docs: audioshake.mintlify.app/api-reference

Endpoint	Method	Description
/assets	POST	Upload a file
/tasks	POST	Create a processing task (up to 20 targets)
/tasks/{id}	GET	Check task status and get results

Output formats: wav, mp3, flac, aiff, mp4, json, srt, txt

Note: Output download links expire after 1 hour. Re-fetch the task to get fresh links.