Remove Music from a Voice Recording
When a voiceover, interview or clip has music underneath, this tool removes the music bed and keeps the speech. The Mel-Band RoFormer engine separates the voice from the backing track.
How it works
Mel-Band RoFormer treats the spoken voice as the "vocal" stem and the music as accompaniment, returning the isolated speech. This rescues dialogue that was mixed over a song or score.
What it's good for
- Clips with copyrighted music beds
- Interviews over background music
- Cleaning music from voiceovers
- Salvaging dialogue from videos
Details
- Engine
- Mel-Band RoFormer
- Formats
- MP3, WAV, M4A, FLAC, OGG, AAC, MP4, MOV
- Price
- Free to try
Frequently asked questions
Same engine, different intent: vocal isolation is framed for songs; this is framed for spoken voice over music. Both return the voice stem; pick the one matching your material.
Speech comes out clearly with the music greatly reduced. Loud music in the same range as the voice can leave faint traces; follow with the background-noise tool if needed.
Very loud music that overlaps the voice's frequencies is the hardest case, but the model still pulls the speech forward in most clips.
No. Removing a music bed only changes the audio; if the clip used copyrighted music, taking it out does not grant you any rights to the original recording.
A clean source where the voice is reasonably forward in the mix works best; you can upload common audio formats, or extract the audio track from a video first.
Short voiceovers and interview clips usually finish in seconds to a minute, scaling with the length of the recording rather than how loud the music is.
Some residue can survive when music sits in the same frequency range as the voice; running the cleaned speech through the background-noise tool afterward smooths over most of it.