All products

Lower music under speech with AI

Use an existing voice recording or video and a separate music bed to keep words audible and bring the music back during pauses.

Lower music under speech with AI when a music bed sounds right during the introduction but masks the presenter’s first words. The audio-ducking skill sends an existing voice input and a separate music track to Sonilo. The service lowers the music during speech and raises it in the gaps.

For an editor working on a short interview, the voice input can be the interview video itself. The result comes back as a new video with the mix attached. A voice recording supplied as audio produces a WAV mix that you can place in your own timeline.

What it does

  • Use a separate music bed beneath an authorised spoken recording.
  • Supply a talking-head clip directly as the voice input.
  • Listen to the transitions between words and short pauses.
  • Receive a new video when the original voice input includes picture.
  • Check the existing processing job if the initial wait times out.

Install and use

  1. Prepare the two files, permission to process them online and official Sonilo access through a connected tool or authenticated CLI. Processing may incur a charge.
  2. Ask, “Mix this approved interview with the instrumental underneath. Return a new video and keep the originals.”
  3. Listen to the start and end of several phrases to check the balance against the source.

FAQ

Source

Made by Sonilo. Original skill: audio-ducking, published on GitHub under the MIT licence.

View on GitHub