Voice isolator: separate speech from noise, music and reverb
Chat shortcut: /voiceisolate
Voice isolator: upload a recording and get a speech track with less background noise. The AI brings the voice forward and dampens traffic, wind, music and reverb. How much remains depends on the recording; a short test is worth it for outdoor interviews, meeting recordings and video sound with music.
With Account
Cost: 26,700 tokens per minute or part thereof · The uploaded file is deleted after 24 hours; the result stays in “My media”.
This tool can be used with a free account. Sign up for free
How the AI separates voice and surroundings
A classic noise filter is designed for steady disturbances such as fans or tape hiss. Traffic, wind, chatter in a café, music in the background or reverb in the room change constantly and are harder for it to grasp. The model behind this tool works differently: it has learned from many recordings what human speech sounds like, lifts the voice out as a separate track and reassembles the recording without the rest.
The result is a speech track in which the voice is in the foreground. What the model does not recognize as voice is dampened or removed, even if you wanted to keep it, such as an intentional music bed. There is no speaker selection: several speakers in a conversation are usually kept, voices in the background may remain or sound altered. Very quiet speech or heavily clipped recordings remain difficult; listen through the result once before using it.
You need the tool for outdoor interviews, meeting recordings, voice notes from your phone and video sound in which music or street noise drowns out the speech. Afterwards the track can be transcribed, captioned or put back into a video.
How it works
- 1
Upload the recording
Drag the audio file into the box or select it: MP3, M4A, WAV, OGG or FLAC up to 50 MB, at least 5 seconds and at most 30 minutes; convert a video's audio first with “Convert MP4 to MP3”.
- 2
Check the cost and start
The price per minute or part thereof is above the form; “Let’s go” starts the separation, there are no settings.
- 3
Listen and download
Listen to the processed track and download it as MP3; afterwards you can transcribe it or create captions from it.
Typical uses
Interviews and reports
Dampen street noise, wind and audience in outdoor recordings before the piece is edited.
Meetings and phone calls
Clean up recordings with keyboard clatter and background conversations so that transcription and captions come out cleaner.
Video sound
Lift speech out of clips with background music, for example to put a new bed under the commentary; music remnants may stay, and the tool does not clear rights to the music.
Good to know
- Input: MP3, M4A, WAV, OGG, FLAC up to 50 MB, at least 5 seconds and at most 30 minutes; output: one MP3 file with the voice brought forward.
- It reduces uneven disturbances such as traffic, wind, chatter, music and reverb as well as steady noise; there are no sliders and no speaker selection, the model decides itself what is voice.
- The model is not built for vocals in music; for that there is “Vocal Isolator”. For steady noise alone, “Remove Background Noise” is often enough.
- Account required. Billed per minute or part thereof at the price above the form; 10 minutes and 5 seconds count as eleven minutes.
- The recording is transmitted to the AI provider for separation; what the provider stores is governed by its terms. The uploaded file is deleted from our servers 24 hours after upload, the audio result stays in your media library.
- In the chat you reach the tool with the shortcut /voiceisolate.
Frequently asked questions
What is the difference from “Remove Background Noise”?
Denoising filters steady noise such as fans or hum. The AI here recognizes the voice itself and also dampens uneven disturbances such as traffic, music, wind or chatter. For a quiet recording with slight noise, denoising, which is free, is enough.
Does it work with vocals or music too?
The model is designed for spoken speech. To lift the singing voice out of a song, use “Vocal Isolator”; for the instrumental, “Vocal Remover”.
What happens to music I want to keep?
It is removed as far as possible, because everything except the voice counts as a disturbance. Put intended music under the processed speech track again afterwards in editing.
Are reverb and echo removed?
Reverb and room echo are among the disturbances the model reduces. In very reverberant rooms, remnants may remain; testing a short excerpt first saves tokens.
Which formats and lengths?
MP3, M4A, WAV, OGG, FLAC up to 50 MB, at least 5 seconds and at most 30 minutes; the result is an MP3. Convert a video's audio first with “Convert MP4 to MP3” (up to 96 MB and 30 minutes); the resulting audio file has to stay under 50 MB.
What does it cost?
A fixed token amount per minute or part thereof, shown above the form; partial minutes count in full. You need an account. If you are unsure, first cut out one minute with “MP3 Cutter” and try it.
What happens to my file?
It is handed to the AI provider for processing and deleted from our servers 24 hours after upload. The result stays in your media library until you delete it.