What does Audio Ducking Tool do?
Audio ducking lowers the music track while the narration track is loud. A sidechain compressor listens to the voice and turns the music down. It does not isolate a voice from a finished mix, and it is not AI. Threshold is the voice level that starts the reduction. Duck amount is the compressor ratio. Attack is how quickly the music falls, and release is how quickly it returns. Music and voice volumes are applied before the compressor. Light, Balanced and Strong are starting points, not loudness targets.
How to use Audio Ducking Tool
- 1Upload the music and the separate narration file.
- 2Start from Light, Balanced or Strong, then adjust threshold, duck amount, attack and release.
- 3Set the music and voice volumes and listen to both files before export.
- 4Export WAV, MP3 or M4A. Cancel stops the local job.
Supported inputs and outputs
Inputs
Background music, Voice or narration audio
Outputs
WAV, MP3 or M4A with the ducked music mixed back with the voice
Practical uses
- Keeping music under a voiceover
- Trying a stronger duck without rewriting the edit
- Exporting a local mix as WAV, MP3 or M4A
Browser and media limitations
- Ducking needs two tracks. It cannot pull a voice out of a file that already mixes voice and music.
- Fast attack and a high ratio can pump. A slow release can keep the music quiet after the voice stops.
- The mix is re-encoded. WAV is uncompressed PCM. MP3 and M4A are lossy.
- This page does not duck music under a video. Add the finished audio to a video with Add Audio to Video.
How local processing works
Your media is processed locally in your browser and is not uploaded to Tubelexity for processing. The media engine downloads on first export. Page navigation and analytics still use network requests.
Frequently asked questions
Does this separate voices from music?
No. It lowers one file while another file is loud. A voice already mixed into the music cannot be isolated.
What do threshold, attack and release do?
Threshold sets how loud the voice must be before the music drops. Attack sets how fast that drop happens. Release sets how fast the music comes back.
Is this AI?
No. The reduction is sidechain compression, a conventional dynamics process.
Are the recordings uploaded?
No. Both files stay in this browser. The media engine downloads on first export.