Recut
How to remove background noise from a video
Attach the video and ask Recut to "remove the background noise". It takes the sound out of the video, runs a voice-isolation model to separate the speech from noise and music, and puts the clean voice back on the original picture.
What to type
Clean up the audio in this video: keep my voice and remove the fan noise.
Step by step
- Attach the videoAdd the noisy recording to a Recut project.
- Ask for clean audioSay what to keep and what to lose: "keep the voice, remove the traffic".
- Check the planRecut plans three steps: take out the sound (on your device), isolate the voice (a model run, priced before you start), and put the voice back on the video with the old sound muted (on your device).
- Run and comparePlay the original and the cleaned version side by side on the timeline.
What Recut does
Audio models work on sound files, so Recut first saves the video's sound as an MP3. It then suggests separation models, such as ElevenLabs' audio isolation, Demucs and SAM Audio, which split a recording into the voice and everything else.
Finally the clean voice is added back to the video with the video's own sound set to zero. The picture is copied untouched, so only the sound changes.
Tips
- Isolation removes music under the voice too. If you want music back, add it as a separate track at a low volume afterwards.
- After cleaning, an export for your platform brings the voice up to a steady loudness.
What it doesn't do
- Results depend on the recording: a voice buried under very loud noise may come out thin or with artefacts.
- There is no adjustable noise-reduction slider; the model decides what is voice.
Try it on your own video
Recut runs in your browser, on Mac, Windows and Android. Start free, attach a file and say what you want.
Open Recut in your browserRelated guides
- How to separate vocals from music in a songRecut runs a stem separation model, such as Demucs or SAM Audio, to split a song into vocals and instruments, or remove the vocals for karaoke.
- How to extract the audio from a video (MP3 or WAV)Ask Recut to save a video's sound as an MP3 or WAV. It runs on your device with ffmpeg, free of model costs, and keeps the original video.
- How to normalize a video's audio loudnessRecut measures a video's loudness and sets it to the level TikTok, Reels, YouTube or LinkedIn play at (-14 or -16 LUFS), with a limiter on peaks.
- How to transcribe a video to textAttach a video and ask for a transcript: Recut takes out the sound and runs a speech-to-text model such as ElevenLabs Scribe or Whisper.