Mureka AI Stem Song API Documentation

Mureka AI Stem Song API Documentation

Playground

Mureka AI Stem Song is a fast AI music model that creates stem song outputs. Ready-to-use REST inference API for AI music generation, vocal and instrumental stem workflows, remixing, music production, creative audio projects, and professional song creation with simple integration, no coldstarts, and affordable pricing.

Features

Mureka AI Stem Song is a fast AI music model for stem-based audio workflows. Upload a song or audio track, choose a separation model, and generate stem outputs for remixing, vocal/instrumental workflows, editing, and music production tasks.

Why Choose This?

• Fast stem generation Separate an input track into usable stem outputs for remixing, editing, and production workflows.

• Multiple separation models Choose between available stem models such as audio-separation-1 and audio-separation-2.

• Simple audio workflow Upload one audio file, choose a model, and generate the separated output with minimal setup.

• Useful for production and editing Suitable for vocal isolation, instrumental extraction, remix preparation, and creative audio workflows.

• Production-ready API Easy to integrate into music tooling, audio apps, and content-production pipelines.

Parameters

How to Use

1. Upload your audio — provide the source song or audio file you want to process.
2. Choose the model — select the separation model that best fits your workflow.
3. Submit — run the model and download the generated stem output.

Example Use Case

Upload a full song to separate vocals and instrumental layers for remixing, re-editing, or production cleanup.

Pricing

Audio Separation V1 costs $0.09 per song; V2 costs $1.05 per song.

Best Use Cases

• Stem separation — Extract parts of a track for editing and reuse.
• Music production — Prepare vocals or instrumentals for remixing and arrangement work.
• Content creation — Generate cleaner audio components for videos, shorts, and social content.
• Remix workflows — Isolate audio elements for mashups, reinterpretations, and creative edits.
• Audio cleanup and reuse — Separate usable parts of an existing recording for downstream production.

Pro Tips

• Use the cleanest source audio available for better stem quality.
• Try both available models if you want to compare separation behavior.
• Full mixes with clear vocals and instrumentation often produce more usable stems.
• Review the separated outputs before using them in final production workflows.

Notes

• audio is required.
• model is optional.
• Pricing is $0.09 for Audio Separation V1 and $1.05 for V2.

Related Models

• Other Mureka AI music generation and audio processing models may be useful when you need different music or audio workflows.

Authentication

For authentication details, please refer to the Authentication Guide.

API Endpoints

Submit Task & Query Result

set -euo pipefail

export WAVESPEED_API_KEY="your-api-key"

REQUEST_BODY=$(cat <<'JSON'
{
  "audio": "https://interactive-examples.mdn.mozilla.net/media/cc0-audio/t-rex-roar.mp3",
  "model": "audio-separation-2"
}
JSON
)

# 1. Submit the prediction once. Never retry this request automatically:
# a retry after an unclear failure can create a second, billed task.
CHECK='The task may already exist. Check History or the predictions list before submitting again.'
if ! SUBMIT_RESPONSE=$(curl --silent --show-error --fail-with-body \
  --connect-timeout 10 --max-time 60 \
  -X POST "https://api.wavespeed.ai/api/v3/mureka-ai/stem-song" \
  -H "Authorization: Bearer ${WAVESPEED_API_KEY}" \
  -H "Content-Type: application/json" \
  -d "${REQUEST_BODY}"); then
  printf 'Submit failed. %s\n' "${CHECK}" >&2
  exit 1
fi

TASK=$(printf '%s' "${SUBMIT_RESPONSE}" | jq 'if type == "object" and has("data") then .data else . end')
PREDICTION_ID=$(printf '%s' "${TASK}" | jq -r '.id // empty')
if [ -z "${PREDICTION_ID}" ]; then
  printf 'Submission response did not contain a prediction id. %s\n' "${CHECK}" >&2
  exit 1
fi
RESULT_URL="https://api.wavespeed.ai/api/v3/predictions/${PREDICTION_ID}/result"

# 2. Poll until the prediction reaches a final status.
BODY_FILE=$(mktemp)
trap 'rm -f "${BODY_FILE}"' EXIT
while true; do
  # Transient errors (429, 5xx) are retried, honoring Retry-After.
  HTTP_CODE=$(curl --silent --show-error --fail-with-body \
    --connect-timeout 10 --max-time 60 --retry 5 --retry-delay 2 \
    --output "${BODY_FILE}" --write-out '%{http_code}' \
    "${RESULT_URL}" \
    -H "Authorization: Bearer ${WAVESPEED_API_KEY}" || true)
  case "${HTTP_CODE}" in
    2??) ;;
    000|429|5??)  # dropped connection or temporary error: keep polling
      printf 'Task %s: HTTP %s, polling again in 10s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      sleep 10; continue ;;
    *) printf 'Task %s: HTTP %s\n' "${PREDICTION_ID}" "${HTTP_CODE}" >&2
      cat "${BODY_FILE}" >&2; exit 1 ;;
  esac
  RESULT=$(jq 'if type == "object" and has("data") then .data else . end' "${BODY_FILE}")
  STATUS=$(printf '%s' "${RESULT}" | jq -r '.status // empty')

  case "${STATUS}" in
    completed) printf '%s\n' "${RESULT}" | jq '.outputs'; break ;;
    failed|cancelled|timeout|deleted)
      printf 'Task %s ended with status %s\n' "${PREDICTION_ID}" "${STATUS}" >&2
      printf '%s\n' "${RESULT}" | jq . >&2; exit 1 ;;
    *) sleep 2 ;;
  esac
done

Parameters

Task Submission Parameters

Request Parameters

Every model also accepts the global request options enable_sync_mode and enable_base64_output; see their pages for details.

Response Parameters

Result Request Parameters

Result Response Parameters