Sonilo / Voice-over

Music That Leaves Room for Voice-Over

Upload your video with the narration in place, and Sonilo composes a track that thins out under speech and builds where the visuals carry the moment.

Start free with limited credits. Free and Creator output is for personal, non-commercial use.

  • +24.35%average lift in video engagementSonilo's internal usage data, versus picking a track from a music library, as of September 2026
  • 5.3Mcopyright-risky soundtracks replacedSonilo generations to date, as of September 2026
  • 700+brands create with Sonilo, from indie studios to global teamsaccounts with a paid plan, as of September 2026
  • Licensed catalogthe model is trained on a licensed Shutterstock catalog, not scraped audiosee /licensing

How it works

How Sonilo composes around a voice-over

  1. 01

    Upload the video with your voice-over already placed

    Sonilo reads the visuals and the pacing of the edit; the original audio, including the narration, is kept.

  2. 02

    Enable Keep speech and vocals, then choose a mood

    Add "under narration" or "documentary bed" to the prompt if you want extra restraint.

  3. 03

    Generate options

    Each option is arranged so the melodic and rhythmic density drops in talking sections and rises in B-roll and transitions.

  4. 04

    Export

    Export with the music mixed under your narration at the level you set, or download the music as its own file and mix it in your editor.

Who this is for

Best for: ads, documentaries, explainers, and any video where someone is talking

Stock tracks are written to be heard on their own, so editors spend time ducking and cutting them under speech. Sonilo composes the track for the edit it is given, with speech space treated as part of the arrangement rather than a volume automation added afterward.

Turning a track down keeps its melody fighting the voice. Sonilo changes what is played, not only how loud it is.

If the voice and music already exist as separate files and only the balance needs fixing, the audio ducking workflow does that without generating new music.

Before you publish

Licensed source. Plan-based rights.

The model is trained on a licensed Shutterstock catalog. Free and Creator output is for personal, non-commercial use. Pro, Premium, Enterprise, and eligible API users have commercial-use rights under the current Sonilo Terms.

For monetized channels, paid ads, client projects, or a shipped game, check the plan used for generation and keep the output and license records with the project.

Common questions

Music That Leaves Room for Voice-Over, answered.

How does Sonilo know where the speech is?

With Keep speech and vocals enabled, Sonilo treats the uploaded audio as something the music has to sit under and uses the visuals and pacing of the edit, such as talking-head framing and scene changes, to shape the arrangement. Review the result against the picture; generated output still needs editorial review.

Can I still duck the music manually?

Yes. Download the music as a separate file and mix it in Premiere Pro, DaVinci Resolve, or CapCut. The editor guides in the AI Music hub cover each one.

Is this different from a low-volume stock track?

Yes. A quieter stock track still carries a full melody under the voice. Sonilo composes a sparser arrangement in the talking sections, so there is less to compete with.

Can I use the music in client work or paid ads?

Free and Creator output is for personal, non-commercial use. Pro, Premium, Enterprise, and eligible API users have commercial-use rights under the current Sonilo Terms. Check the plan used for the generation and the intended release before publishing.