Skip to content

VOD AIGC Update: EL compose v2.5 Music Generation and sound_t2s_v3 Sound Effect Generation Models Released ​

Updated

VOD AIGC adds three versions to the EL (ElevenLabs) model (released 2026-11-16). The action is still CreateAigcAudioTask; SceneType is not specified.

ModelNameModelVersionPurposeKey capabilities
ELcompose_v2.5MusicPrompt to music
ELcompose_detailed_v2.5MusicPrompt to music (ElevenLabs "Compose music with details" endpoint)
ELsound_t2s_v3Sound effectsText to sound effect

1. Music: compose v2.5 ​

  • Use compose_v2.5 for general music generation; the existing compose_v2 remains available.
  • OutputConfig.Duration does not apply to the compose_* music versions.
  • ⚠️ Whether compose_detailed_v2.5 accepts a composition_plan (section-level structure) and returns extra metadata is not yet confirmed.

2. Sound effects: sound_t2s_v3 ​

  • For short sound effects, the latest sound_t2s_v3 is recommended; the existing sound_t2s_v2 remains available.
  • sound_t2s_v3 takes the same parameters as sound_t2s_v2. Set the length with OutputConfig.Duration (seconds, [1, 30]).
  • To migrate an existing v2 request, just change ModelVersion to sound_t2s_v3.

3. Docs ​

Built with VitePress · Deployed on Tencent EdgeOne Pages