---
title: "Generate audio using AI"
method: POST
path: "/collection/{collection_id}/generate/audio/"
---

# Generate audio using AI

`POST /collection/{collection_id}/generate/audio/`

## Path parameters

- `collection_id` string, required

## Request body

- object
  - `prompt` string — Prompt for music or sound effect generation.
  - `text` string — Text to convert to speech when `audio_type` is `voice`.
  - `audio_type` 'voice' | 'sound_effect' | 'music', required
  - `voice_name` string — Voice name for hosted text-to-speech.
  - `model_name` string — Model used for text-to-speech.
  - `config` object — Voice configuration. For OmniVoice supports instructions, ref_audio, ref_text, response_format, speed, language, and max_new_tokens.
  - `sandbox_id` string — Sandbox ID used for self-hosted text-to-speech.
  - `voice_clone_id` string — Reusable VideoDB voice clone ID.
  - `callback_url` string

## Response `200`

Audio generation started

- AsyncResponse
  - `success` boolean
  - `status` 'processing' | 'done' | 'failed'
  - `data` object
    - `id` string
    - `output_url` string

---

[API](https://skmtc.net/video-db/apis/videodb-server-api.md) · [All operations](https://skmtc.net/video-db/apis/videodb-server-api/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/video-db/videodb-server-api/versions/21b1d1d3eae8/schema)