---
title: "Clean voice from background noise"
method: POST
path: "/api/v1/split/voice_clean/"
tags: ["Stem Separation"]
---

# Clean voice from background noise

`POST /api/v1/split/voice_clean/`

Typical usage case - is voice with background noise.
    Voice clean will separate clean voice (stem) and noise (back).
    Result of /check/ includes 2 tracks:
    - stem:{"type":"stem", "label":"voice", "url":"..."}
    - back:{"type":"back", "label":"no_voice", "url":"..."}

## Request body

- VoiceCleanSplitParameters
  - `source_id` string, uuid4, required — ID of the source file to be processed.
  - `presets` VoiceCleanPresetsV1, required
    - `splitter` 'andromeda' | 'perseus' | 'orion' | 'phoenix' | 'lyra' | 'lynx'
    - `dereverb_enabled` boolean — Enabling dereverb removes echo from the audio for clarity, which may slightly alter the voice. Disabling dereverb preserves the natural echo of the recording for authenticity. Only for {'vocals', 'voice'} stems.
    - `encoder_format` 'mp3' | 'wav' | 'flac' | 'aac' | 'ogg'
    - `stem` 'voice', required — Stem to extract. Only 'voice' stem is supported.
    - `noise_cancelling_level` 0 | 1 | 2
  - `idempotency_key` string, uuid4, nullable — Unique uuid4 key to ensure idempotent requests. Prevents duplicate task execution with the same parameters. Can be reused until the task starts. Once processing begins, reusing this key will return an error.

## Response `200`

OK

- Task
  - `task_id` string, required — ID of the created split task.

## Other responses

- `400` — Bad Request
- `422` — Unprocessable Entity

---

[API](https://skmtc.net/lalal/apis/lalal-ai-api.md) · [All operations](https://skmtc.net/lalal/apis/lalal-ai-api/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/lalal/lalal-ai-api/revisions/ebb545e26e94/schema)