Voice Extraction API

REST · prepaid credits · no subscription

The same Demucs (htdemucs) voice extraction that powers this site, as a simple REST API. Upload audio, get back the isolated voice with music and background sound removed. Built for transcription pipelines, podcast tools, archive cleanup, and voice datasets.

Pricing

PackPriceMinutesPer minute
Starter$10250$0.040
Plus$401,100$0.036
Scale$1505,000$0.030

Credits never expire. Billing is per minute of audio processed (rounded up to 0.01 min). Failed jobs are automatically refunded. Comparable services charge $0.05–0.07/min.

Starter — $10

Plus — $40

Scale — $150

Authentication

Send your key as a bearer token on every request:

Authorization: Bearer sk-vox-...

Endpoints

1. Submit audio

curl -X POST https://acapellize.com/v1/separate \
  -H "Authorization: Bearer $KEY" \
  -F "[email protected]"

# → {"id":"a1b2c3d4e5f6","minutes_charged":3.51,
#    "balance_minutes":246.49,
#    "status_url":"https://acapellize.com/v1/jobs/a1b2c3d4e5f6"}

Formats: mp3, wav, flac, m4a, ogg, opus, aac, wma. Max 40 minutes of audio; max 100 MB per request.

2. Poll status

curl https://acapellize.com/v1/jobs/a1b2c3d4e5f6 \
  -H "Authorization: Bearer $KEY"

# → {"id":"a1b2c3d4e5f6","status":"done","error":null,
#    "minutes_charged":3.51,
#    "vocals_url":"https://acapellize.com/v1/jobs/a1b2c3d4e5f6/vocals"}

Status is queued → processing → done (or error, which auto-refunds). Typical processing time: 10–30 s for a 4-minute file. Results are deleted after 2 hours — download promptly.

3. Download the voice

curl -o vocals.mp3 \
  https://acapellize.com/v1/jobs/a1b2c3d4e5f6/vocals \
  -H "Authorization: Bearer $KEY"

Check balance

curl https://acapellize.com/v1/balance -H "Authorization: Bearer $KEY"
# → {"balance_minutes":246.49}

Notes