# Noise Removal & Voice Isolation

Remove background noise and music from speech: send audio or video, get the clean voice back as a 48 kHz MP3.

Send a direct link to an audio or video file, or the file itself. You get the speaking voice alone as a mono 48 kHz MP3, with background noise and music removed, within 24 hours and usually within minutes. Ask for the background too and you also get what was taken out. Pay per second of audio; if we miss the deadline it's free. Recordings run on our own hardware in the EU and are never used for training.

- Noise reduction made for speech: interviews, calls, podcasts and voice notes come back as the speaking voice alone, with noise and music removed. It is an audio denoise model, not a filter.
- Ask for the background too and you also get what was taken out.
- Pay per second of audio. Recordings run on our own hardware in the EU and are never used for training.

## Calls

### `POST https://readymadeapi.com/v1/audio/clean`

Queue a clean-up. max_minutes is the most the audio may last; you are charged only for the seconds it actually runs.

- Price: $0.00005 per second, up to 3600 per call
- MCP tool: `audio_cleanup_clean`
- Queued: answers 202 with a job; poll `GET /v1/jobs/{id}`, fetch `GET /v1/jobs/{id}/result`. Charged up front, and what does not run is refunded.

Request:

```json
{
  "audio_url": "https://readymadeapi.com/samples/voice-note.mp3",
  "max_minutes": 30
}
```

The finished job:

```json
{
  "job_id": "01a1f552-fffa-72ff-b38e-010d24d80112",
  "status": "succeeded",
  "late": false,
  "deadline": "2026-11-20T19:40:19.823506Z",
  "completed_at": "2026-11-19T19:40:21.58172Z",
  "error": null,
  "result": {
    "audio_mime": "audio/mpeg",
    "duration_s": 6.525,
    "audio_base64": "SUQzBAAAAAAAIlRTU0UAAAAOAAADTGF2ZjYyLjMuMTAwAAAAAAAAAAAAAAD/... (105276 characters)"
  }
}
```

## Limits

- Up to 60 minutes of audio per call.
- By link, any common audio or video format, up to 350 MB.
- The result is mono.

## Questions

**What is max_minutes for?**

It is your ceiling. Audio longer than that is refused and not charged; shorter audio is charged for its real length.

**How long does it take?**

It is a queued job: you get a job id at once and poll for the result. Most files finish within minutes; the promise is 24 hours, and if we miss it the job is free.

**Does it split music into stems?**

No. It separates the speaking voice from everything else. It is not a vocals, drums and bass splitter for songs.

**What can I send?**

A direct link to an audio or video file, or the file itself as base64 when it fits in 2 MB. Up to one hour per call.

## Using it

- Send `Authorization: Bearer <key>` and a JSON body. Keys and credit: https://readymadeapi.com/account
- Every answer carries `X-Request-Id`, `X-Units` and `X-Charged-Usd`.
- An error is a JSON problem with a `title` to act on and a `detail` in plain words; nothing is charged for it.
- Full schema: https://readymadeapi.com/openapi.json
