Introduction - AudioShake Developers

Your first API call

Separate a track into vocals and instrumental in a single request:

curl -X POST "https://api.audioshake.ai/tasks" \
  -H "Content-Type: application/json" \
  -H "x-api-key: your_api_key" \
  -d '{
    "url": "https://demos.audioshake.ai/demo-assets/shakeitup.mp3",
    "targets": [
      { "model": "vocals", "formats": ["wav"] },
      { "model": "instrumental", "formats": ["wav"] }
    ]
  }'

What you can build

• Remove vocals from songs
  Isolate vocals, drums, bass, guitar, and more for remixing, karaoke, and sampling.

• Isolate dialogue for dubbing
  Separate speech from music and effects for localization and post-production.

• Transcribe and sync lyrics
  Generate word- and line-level timestamped lyrics from any song.

• Separate speakers
  Isolate individual speakers from multi-speaker recordings, even with overlapping speech.

• Detect and identify music
  Find where music appears in podcasts, video, and broadcast content for compliance.

• Recover degraded speech
  Denoise and de-reverb recordings for clearer speech and better transcription.

Who this is for

• AI companies training speech, music, or multimodal models with clean, labeled data
• Media companies processing archives for dubbing, compliance, and cataloging
• Developers building karaoke apps, remix tools, practice platforms, and audio experiences

Why AudioShake

• State-of-the-art quality — purpose-built models for music and speech, trained on licensed data
• Production-ready — async processing, webhooks, and batch support for any scale
• Simple API — one endpoint, multiple models per request, results in minutes
• Audio + video — process MP4 and MOV files directly, no pre-processing needed
• On-device option — run the same models locally with the Local Inference SDK

Reference

• Models
  Browse all available models with pricing.
• API Reference
  Full endpoint reference.
• Tutorials
  Step-by-step guides for every workflow.
• Billing & Credits
  How pricing works.