Video to Music
By Steven Van ·
The new API scores up to 10 videos (200 MB combined) as one, with optional sound effects and separate music and SFX tracks.
ElevenLabs has opened a Video to Music API that generates a soundtrack from video files, so a music prompt does not have to be written from scratch. The vendor suggests it for product videos, ads and social clips.
Input limits
A request can include up to 10 video files, with a combined size of 200 MB and a total length of up to 600 seconds. The files are joined in the order sent and scored as one video. The documentation notes that combining several videos can lengthen the request noticeably, and recommends merging them beforehand where possible.
Steering the result
- An optional text description of up to 4,100 characters and up to 10 style tags (for example "upbeat" or "cinematic") influence the music.
- use_audio_as_reference takes the genre, BPM, instrumentation, mood and arc from the video's own audio track.
- add_sfx adds sound effects that match cues in the video.
- separate_tracks returns the music and the sound effects as separate files in a ZIP, so they can be mixed and layered separately.
Models and output
The model_id parameter accepts music_v1 (the default), music_v2 and music_v2_5. The output defaults to MP3, and the output_format parameter selects other formats. MP3 at 192kbps requires the Creator tier or above, and PCM at 44.1kHz requires Pro or above. Generated MP3 files can optionally be signed with C2PA using sign_with_c2pa. Each response includes a song-id header identifying the generated song.
