ElevenLabs Instant Soundtracks scores a video with music and sound effects from one upload

ElevenLabs launched Instant Soundtracks on October 15, 2026, and the pitch fits in one line: upload a video, and ElevenCreative scores it with original music, then watches the footage for moments that need a sound effect and drops a generated one at the right timestamp. ElevenLabs’ own example is a whoosh timed to the logo at the end of an ad.

It also runs on the ElevenMusic platform and through the Video to Music API, which takes up to 10 minutes of footage per request.

One honest caveat before the details: I haven’t run Instant Soundtracks on my own footage yet. Everything below comes from ElevenLabs’ launch post and its API documentation, not from a test.

How ElevenLabs Instant Soundtracks scores a video, step by step

Instant Soundtracks does 2 jobs on one upload, which ElevenLabs sells as a single pass. First the model writes an original score for the whole video. Then it goes back through the footage, finds the moments that call for an effect, and generates one for each, placed at its timestamp.

If your clip already has audio, you can hand that over as a reference. ElevenLabs says the model reads its genre, crescendos, beats per minute and instrumentation, and shapes the new score to match the original’s character instead of starting from scratch.

The timing is tuned for short-form video, with tighter hits around cuts, transitions and on-screen actions, and the mood can change between scenes. ElevenLabs’ example: a fast scene gets EDM or R&B, then the track moves to acoustic guitar when the scene slows down.

The part that matters most for editing is the layout. In ElevenCreative’s Music Video Editor, the sound effects sit on their own layer, separate from the soundtrack. You can add more effects, move their timing, swap the music or fix a single moment without regenerating everything, then export straight from ElevenCreative.

If you’d rather wire voice, sound effects and music together by hand, ElevenLabs Flows does that on a node canvas. Instant Soundtracks is the one-button version, and under the hood it’s also an API call.

The ElevenLabs Video to Music API: add_sfx, separate_tracks and use_audio_as_reference

The Video to Music endpoint generates background music from one or more videos. ElevenLabs’ changelog for October 19, 2026 added 3 optional switches, all off by default:

• add_sfx adds sound effects cued from the video.
• separate_tracks returns a ZIP with the music and the sound effects as 2 separate files instead of one mixed file, so you can balance them yourself in your own pipeline or in Studio.
• use_audio_as_reference analyzes the video’s existing audio and uses its genres, BPM, instrumentation, mood and arc as the reference for the new music.

The videos are joined in order and scored as one video, and ElevenLabs warns that combining several can slow the request down significantly, so stitch them first if you can. A text description of up to 4,100 characters and up to 10 style tags steer the result.

Two more details from the reference. The endpoint needs a paid plan, because it returns a subscription-required error without one. And an optional sign_with_c2pa flag signs MP3 output with C2PA provenance data, which records where the file came from.

What an AI soundtrack generator for video does, and what it doesn’t

An AI soundtrack generator for video writes music to picture. You give it footage, and it returns a score timed to that footage, so the music moves with the edit instead of running underneath it like a library track you trimmed to fit. Instant Soundtracks adds the sound design on top: the whooshes and hits an editor would otherwise place by hand.

What it doesn’t do is write your song. The brief is the footage, and the job is a score for one clip (for longer scoring work, my ElevenMusic film scoring guide covers the workflow).

ElevenLabs says Eleven Music was created in collaboration with labels, publishers and artists, and it launched in August 2025 with licensing deals with Merlin and Kobalt (I covered the latest model in my Music v2.5 piece). That’s ElevenLabs’ claim about where the model comes from. What you’re allowed to do with the output is in its Music Terms, last updated October 9, 2026:

• Every self-serve plan, from Free to Business, allows online and offline commercial use except film, TV, radio and Studio Games. Those need the Enterprise Music plan.
• The Free plan can’t download music, and it requires you to credit Eleven Music when you distribute.
• Music streaming distribution is prohibited on Free and Starter, and permitted from Creator up.
• Free through Pro are for individual use only.

So a Reel, a YouTube ad or a product video on a paid plan is covered. A TV spot isn’t, unless you’re on an enterprise agreement.