Skip to main content
Voice management creates reusable voice IDs for supported Kling video generation requests. A custom voice can be created from a clean audio/video URL or from an eligible historical generated video.

​
Source requirements

  • voice_url supports MP3, WAV, MP4, and MOV.
  • Use a clean recording containing exactly one human voice.
  • Source duration must be 5–30 seconds.
  • Alternatively, provide an eligible video_id from a Kling 2.6 task with sound enabled, an Avatar task, or a Lip-Sync task.

API Reference

Create, query, list, and delete custom voices, or list official preset voices.