Isolate one voice
from any recording

Free speaker isolation. Upload an audio or video file and mark a moment where only the voice you want is talking. You get that one speaker alone. Everyone else drops away. Up to 60 seconds, no account.

Try enhancing

Drop an audio or video file here

WAV, MP3, M4A, FLAC, MP4 or MOV. We use the first 60 seconds.

This is not noise removal

We wrote about the difference at length. Editing a video? How to remove one person's voice from a video, step by step.

Cleanup tools separate speech from music and background noise. To them every voice is speech, so the second person in the room stays.

This one builds a voiceprint from a short sample and keeps only that person. Everyone else is removed the way a hum would be. The technical name is target speaker extraction.

You still get the cleanup. Everything that is not your speaker goes: the other voices, and the hum and traffic too. If what you want is one voice made to sound like a studio recording, the enhancement tool does that.

Diarization labels who spoke when. Separation splits everyone onto their own unlabeled track. Neither returns the thing you actually want: this speaker, and nobody else.

Questions

What counts as a good sample?
Three seconds or more of your speaker talking alone. Clean matters more than long: a sample with two voices in it keeps both.
Is it really free?
Yes, free for recordings up to 60 seconds, no account needed. If you need to process longer files, join our waitlist.
What happens to my audio?
The result link stops working after an hour. For more on how we handle your files, see the privacy page.
Why does the result sound different from my upload?
The output is tuned for speech, so it can sound a little narrower than the original. The before and after are matched, so the difference you hear is the separation.
Can it make the isolated voice sound better?
Isolation keeps your speaker and drops everything else, but it hands back the voice as it was recorded, room and all. The speech enhancement tool on this site takes it from there: give it the isolated track and it regenerates that one voice clean, without the room it was recorded in.
Can I get more than one speaker?
Each run returns one speaker, the one you sampled. If you need every voice on its own track, join our waitlist and tell us.
Is this the real-time engine?
No, this is the free batch tool for recorded files. The real-time engine is coming soon. If you want early access to it, join our waitlist.
Can you remove one person's voice from a recording?
Yes. Pick the voice you want to keep, and every other voice is removed along with the background. Forum answers that call this impossible are thinking of equalizers and phase tricks, which work on frequencies. This works on identity: it learns one voice and keeps only that.
Why doesn't a noise remover take out the other voice?
Because to a noise remover, every voice is the thing worth protecting. It strips hum, traffic, and music and keeps all speech, including the person you wanted gone. Removing a specific voice needs a tool that knows which speaker is yours. That is what this one does.
Does it also remove background noise?
Yes. Everything that is not your speaker is removed, the background noise along with the other voices. The difference from a noise remover is the direction: a noise remover keeps every voice and drops the noise, while this keeps one voice and drops everything else.
Can you separate two people talking at the same time?
Yes. Overlap is the normal case here, not the failure case. The tool follows your speaker through the crosstalk and returns their side alone. The one requirement is the sample: mark a moment where your speaker is talking by themselves, not a moment where both are going at once.
Does it work on phone calls and voice memos?
Yes. A call recording, a voice memo that caught the whole room, an interview, a lecture: any audio or video file with your speaker somewhere inside it. WAV, MP3, M4A, FLAC, MP4 and MOV all work. This tool is built for rough recordings.
Can I remove background conversations from a video?
Yes. Drop the video on the tool, MP4 or MOV. Your browser reads the audio track out of the file, so only the audio is sent, and the voice you mark comes back on its own as a clean audio file to lay back over the picture.
Do I need an audio editor like Audacity or Premiere?
No. Waveform editors operate on the mixture, so their vocal-isolation effects can duck music and noise but cannot take one voice out of another. This runs in the browser: upload, mark your speaker, download the result. No install, no plugins, no account.
How is this different from a vocal remover?
A vocal remover splits singing from instruments, which is a music problem. This is a speech problem: several people talking, and you want one of them. It picks a single speaker out of the other voices, which is the part vocal removers and karaoke tools do not attempt.
Can I isolate my own voice from a recording?
Yes. You are just another target speaker. Mark a moment where you are talking alone, or upload a short separate sample of your voice, and the result is your side of the recording with everyone else removed.
What if the result still has traces of the other voice?
Re-run it with a better sample. Almost every weak result traces back to enrollment: the sample was too short, or a second voice sat inside it. Find a cleaner three seconds of your speaker alone. That usually fixes it without touching the recording itself.
Does it work when the two voices sound similar?
Similar voices are the hard case, and the sample matters more there. Give it a longer stretch of your speaker alone, ideally somewhere their voice is clear and close to the microphone. The more distinct the voiceprint, the cleaner the separation.