To reuse an interview or voiceover that already contains music, separate its soundtrack before editing the speech. Start with General audio for a mixed interview recording, narrated product demo, or tutorial. Listen to the separated voice, compare it with the source, and keep the original available throughout the edit.
Is this recording suitable for music separation?
This workflow is useful when speech and a music bed are already combined in one audio track. If you have the original editing project with separate microphone and music tracks, mute or adjust the music there first: you already have the independent speech recording.
Choose a source in which the words are audible before processing. A music remover cannot reliably reconstruct words that were never recorded clearly. Interview crosstalk, clipping, wind, and room echo are different problems from a music bed; they require their own review and may need another editing tool or a better source recording.
Use a supported local audio or video file up to 100 MiB, or a public HTTPS link directly to a supported media file. A private share page or streaming watch page is not a direct media link. See the upload and file-handling guide before submitting confidential interview material: processing uses an external service.
How do you prepare interview or voiceover audio?
- Mark reference passages. Listen to the original and note a quiet sentence, a sentence over louder music, and a change of speaker or narration pause. These are your comparison points.
- Open the tool and select the source. Upload your file or enter its direct media link, then sign in to start processing.
- Choose the scene that matches the recording. General audio · Audio produces voice and background tracks for speech recordings. Music produces vocal and accompaniment tracks for songs. The scene describes the source; it is not a quality level.
- Select an audio export format. AAC, MP3, WAV, M4A, FLAC are available. Choose one your editor accepts. WAV or FLAC does not recover detail lost in a compressed or distorted source.
- Review the output before editing. Compare your reference passages, including the ends of words and pauses. Listen to the other tracks too, to check whether needed speech was separated away from the voice.
Keep the page open until the upload finishes and the task is queued. Processing then continues in the background; return to My processed files to check the result.
For a narrated product video, also decide whether clicks, demonstrations, or other audible actions need to remain in the final edit. The voice-only selection uses the voice track. Select background too if you want to include some of the separated non-vocal sound, and adjust its volume before downloading the video.
Which result belongs in your next edit?
| Your next step | Result to use |
|---|---|
| Cut or rearrange spoken answers in an audio editor | Voice audio in a supported export format |
| Reuse the source picture with its separated speech | Voice-only MP4 from a video input |
Rebuild a narrated video's mix with selected effects | Voice and background tracks, mixed in your editor; the background includes music and other non-vocal sounds |
When importing voice audio into the original video project, align it with the source and mute the old mixed soundtrack. Otherwise the old music remains and the speech may play twice. Add your replacement music afterward, adjusting it around the spoken content. If you need the finished picture and voice together, see how voice-only video export works.
What can separation leave behind?
Separation estimates tracks from the mixture. Residual music, softened consonants, or changes to speech can remain, especially during overlapping sounds. Voice separation also does not identify each interview speaker as a separate track, fix microphone clipping, or replace a dedicated echo or noise treatment.
Check the beginning, middle, and end of the saved result in your intended editor. Download the files while their temporary links are available and keep the original. If a word or speaker transition is damaged, use the original passage or another recording rather than assuming the processed track is suitable for the whole interview.
Open the music-removal tool to prepare a source clip, or continue with the
speech-to-subtitles preparation workflow
.