Studio voice AI, one speaker kept
AI Voice Isolator for Noisy Audio and Video
AudioCleaner isolates the main voice from background noise, music beds and other people talking with an AI voice isolator, in audio or video up to 30 minutes and 400 MB. It runs on Studio voice: sign in, see the price, try the first minute. A video’s picture never leaves your device.
- Up to 30 Min and 400 MB
- Try the First Minute
- Background Kept 0 to 50%
- Video Back, Picture Kept
A podcast intro recorded in a café
Café chatter · Podcast · Studio voice
Made noisy on purpose and cleaned on our partner studio, as it came out. The page lines the tracks up so you can switch while it plays.
0:00 / 0:00
Studio voice: the voice rebuilt without the room, so echo and other voices are gone too.
0% is the voice alone. Slide up to bring a little of the place back, for a natural sound. Free to change after the run.
Playback here is matched to it, so the comparison is fair.
Your files and results are visible only to you; runs are kept 7 days.
Up to 30 min · 400 MB
For speech, not song vocals · mono result
- the main speaker kept, the rest removed
- 1 voice
- per file, up to 400 MB
- 30 min
- to try first, never charged twice
- 1 minute
- background kept, set after the run
- 0 to 50%
- MP3 or WAV, mono, loudness presets
- 48 kHz
- results kept, failed runs refunded
- 7 days
Inside the AI Voice Isolator
AI voice isolation for the spoken word: noise, music and other voices out, the speaker kept, and every step easy to check by ear.
Where a Voice Isolator Helps
Six recordings where one voice has to win over everything around it.
Interviews in cafés
Keep the guest; lose the coffee machine, the cups and the next table.
Street vlogs
Traffic, wind and passers-by drop away while the presenter stays in front.
Film and video dialogue
Isolate voices shot on location so the scene can take a new sound bed.
Lectures in a busy room
Chatter and shuffling go, and the lecturer carries the recording.
Podcasts over an intro tune
Pull the host forward from a music bed laid under the talking.
Calls recorded at home
A TV, kids or a dog in the next room: keep the person who was speaking.
How to Isolate Voice From a Recording in 3 Steps
The same steps isolate voice from audio files and videos alike. Sign in first; each run shows its price.
- 01
Open the recording
Drop audio or a video up to 30 minutes and 400 MB: MP3, WAV, M4A, FLAC, OGG, MP4, MOV, MKV and more. It is read in your browser; nothing goes up until you start.
- 02
Try the first minute
Run one minute on Studio voice and switch between the original and the isolated voice while it plays. Wrong speaker kept, or too much gone? You know before paying for more.
- 03
Set the background, then download
Run the whole file, choose how much background to keep and the loudness, then download MP3 or WAV, or your MP4 or MOV with the picture kept.
Voice Isolator Example: A Podcast Intro in a Café
The café sample in the tool above, cleaned on our partner studio and measured by us, before and after.
Café sample, measured by us
Before, the café chatter filled every pause only about 7 dB under the voice, so the host had to compete with the room. After the run, the pauses sit more than 50 dB below the voice and the voice keeps its level. Play the sample and switch views to hear the voice isolator at work.
Voice Isolation, Explained in Plain Words
What does voice isolation mean?
It means keeping one voice and removing everything else in the recording: noise, music, the room and other speakers. In film and video work it is called dialogue isolation; elsewhere, voice extraction. The result is the speaker on a clean, quiet track you can edit, mix or publish.
Voice isolation or noise removal?
Noise removal turns noise down around the voices you recorded and leaves them as they were. Voice isolation rebuilds only the main speaker, so music and other voices go too. Want every voice kept and only the hum or traffic gone? Use the background noise remover, or for wind alone, remove wind noise with its own preset.
What isolates well, and what doesn’t
It works best when the speaker you want is the clearest thing in the file.
- 1Good: one speaker over chatter, traffic, wind, a fan or a music bed.
- 2Harder: two people talking over each other; the louder one is kept.
- 3Not for: songs. Singing over music is a different job.
A Private Voice Isolator: Only the Sound Is Uploaded
Your file opens in your browser and nothing is sent until you start a run. Then only the sound goes up, never a video’s picture. Results wait 7 days for you to download, and you can delete a run from your recent recordings whenever you like.
Voice Isolator FAQ
What voice isolation is, what it costs and what it can and can’t separate.
In plain words, the voice isolation meaning is one voice kept and everything else taken away. Phones and call apps do it live as you talk; a voice isolator like this one does it on a recording afterwards, with time to listen to the whole file, so it can drop music and other voices too.
On a call it is a live mic mode in phones and meeting apps: it filters your voice from the room as you speak, so the other side hears less of your keyboard and street. It changes nothing in a recording you already have; for that, use the voice isolator here.
An AI model trained on a great deal of speech learns what a voice sounds like apart from noise, music and other speakers. Given your file, it predicts the main voice and rebuilds it without the rest. Nothing is cut by frequency, which is why chatter in the same range as the voice can go too.
Yes, that is most of the job: traffic, wind, fans, hum and café chatter go, with music beds and room echo. Want some of the place back? The background slider returns up to half of it after the run.
No. The free quick denoiser in your browser turns down steady noise only, such as hum, hiss and fans; it can’t isolate a voice from music or other people. Isolation runs on Studio voice, billed by the started minute with the total shown first. Try the first minute; a failed run gives the credits back.
Both. Open an MP3, WAV, M4A or FLAC, or an MP4, MOV or MKV: only the sound is uploaded. MP4 and MOV come back as video with only the voice and their picture; MKV, WebM and fragmented MP4s from some phone and screen recorders give the audio.
From a music bed under speech, yes: narration, a vlog with a soundtrack, a podcast with an intro tune. From a song, no. It is built for talking, so singing over music may come out broken. To take the music out from under a whole video, see the background music remover.
It keeps the main speaker, usually the loudest and closest, and removes the others. It does not split each person onto a track of their own. For an interview where both voices matter, pick Clean on this page instead: it turns the noise down and keeps every voice.
No. It runs online in your browser: nothing to install, no editor, no plugin. Open the file, run it, download. For the room ringing on its own, see the echo remover and the reverb remover.
It is billed by the started minute at the Studio voice rate, and the total for your file is on the button before you start. The full run leaves out a minute you already tried, and the same file with the same settings is not charged twice. Plans are on the pricing page.
Audio in MP3, WAV, M4A, AAC, FLAC, OGG, Opus or WebM, and video in MP4, MOV, M4V or MKV, up to 30 minutes and 400 MB. Out comes MP3 (192 kbps) or WAV (16-bit), 48 kHz mono, at podcast, video, audiobook or broadcast loudness. MP4 and MOV can come back from the voice isolator as video.
Try the Voice Isolator on Your Recording
Open an interview, a vlog or a lecture, try the first minute and keep just the speaker. Audio or video up to 30 minutes, the price shown before you start.