Rescroll

Clean up podcast audio, a whole episode at a time.

Drop in an episode, or one host's track, and take out the hum, the hiss and the room around the voices. Any length works: it's cleaned on your device as it's saved, and nothing is uploaded.

  • 100% free
  • No watermark
  • No sign-up
  • Hear it before you save
  • Files stay on your device

Loading the tool…

It runs in your browser. Nothing is uploaded.

How it works

How to remove background noise, free, in three steps.

  1. 1

    Drop in your video or recording

    MP4, MOV or WebM from a phone, a camera or a screen recorder, or an MP3, M4A or WAV. It opens in your browser; nothing is uploaded.

  2. 2

    Hear it before and after

    A stretch of it is cleaned straight away. Switch between before and after, set the strength by ear, and drag to hear another part. The first time, the model downloads once, about 12 MB.

  3. 3

    Download it clean

    The whole file is cleaned on your device as it's saved: the video copied untouched with its new sound, or the recording in the format you choose.

Why this one

Free, with nothing held back, and nothing uploaded.

A neural network, not a filter

DeepFilterNet3, an open-source speech enhancement model, learned what a voice sounds like, so it takes out the noise around one: fans, hum, hiss, traffic, wind, the hubbub of a room. It runs on your own device.

Before and after, on your file

A stretch of your own recording plays cleaned and as it was, switched in an instant, and the strength slider is heard as you move it. The noise floor is measured before and after, in decibels.

Strength you set by ear

Take out everything the model can, or leave a trace of the room so the voice doesn't sound cut off. The strength is a mix of the clean sound and the original, so the voice is never pushed further than you choose.

The picture isn't touched

Only the sound is rewritten. The video is copied into the new file exactly as it was, at its own resolution, with no watermark. Nothing is uploaded, so any length works.

Your file is opened and processed by your own browser. It is never sent to a server, so it stays private, and there’s no upload to wait for.

Cleaning up a podcast's audio

Podcasts are recorded in spare rooms, on laptop microphones and over calls, and a guest's track often carries a fan, a fridge or the street. Cloud cleaners make you upload every episode and usually cap the minutes on a free plan.

Here the episode is cleaned by DeepFilterNet3 in your browser. It reads the file from your disk as it goes and writes the result the same way, so a two-hour episode works like a two-minute one; it just takes longer. A guest's separate track can be cleaned on its own before you mix.

Listen to any stretch before and after first, and set the strength so voices stay natural: a little room sound left in is often better than none. Then save as MP3, M4A or WAV.

Rescroll

Clean sound. Now the edit.

Rescroll's AI video editor edits your talking-head take from a sentence: “cut silences, remove bad takes” trims it, and zooms, a color grade, captions and motion graphics on your words follow by asking. You export a finished MP4.

  • Silences and bad takes cut from one sentence
  • Zooms, captions and a color grade by asking
  • Graphics on your words, designed from a description

Questions

The background noise remover: your questions, answered.

Can it clean a whole episode?

Yes, any length: there's no limit of ours and nothing is uploaded. The sound is cleaned a few seconds ahead of the file being written, so memory never holds the whole episode. On a computer it's much quicker than the episode is long.

Will it remove my intro music?

It can: to the model, music is noise, so music under or between the voices is turned down too. Clean the voice tracks before you add the music, or clean only the spoken part.

How well does it work?

We measured it on the VoiceBank+DEMAND test set, 824 recordings of speech with real recorded background noise under it: at full strength the noise in the pauses came down by 22 dB on average, while the voice changed by about 1 dB. On steady noise, like a fan, hum, hiss or traffic, mixed 10 dB under a voice, it came down by 25 to 38 dB. Your own file's noise floor is measured before and after on the page.

What can't it remove?

Echo: in a room that rings, it turns the tail down only a little (7 to 8 dB in our test) and can make the voice thinner. Rain and clatter it takes down less than steady noise (4 to 11 dB on a recording of rain). Music under a voice counts as noise to it, so a music bed is taken out too. And it can't bring back words a loud noise covered completely.

Will it make my voice sound robotic?

At the default strength a trace of the room is mixed back in, which keeps the voice natural; in our measurements the voice changed by about 1 dB at full strength. If it sounds processed, lower the strength and listen again: before and after switch instantly.

Is it free, and why does it download a model?

It's free, with no account and no limit. The cleaning happens on your device, so the model has to be on your device too: DeepFilterNet3 and its engine, about 12 MB, downloaded from this site the first time and kept by your browser after that. Your file is never sent anywhere.

How long does it take?

Much less than the recording lasts on a computer: the model cleans a minute of sound in a few seconds per channel, and the video isn't re-encoded. A phone takes longer. Keep the tab open while it works.

Is this background noise remover really free?

Yes. There's no price, no trial, no account and no limit on how many files you do. The background noise remover runs on your own device, in your browser, so it costs us nothing to run and we don't charge for it.

Does it add a watermark?

No. The file you download has no watermark, no logo and no intro or outro added. It's your file, changed only in the way you asked.

Is my video or audio uploaded anywhere?

No. The file is opened and processed by your browser, on your device, and never sent to a server: not to us, not to anyone. You can check: watch the Network tab of your browser's developer tools while it works, and you'll see the file is never sent anywhere.

How big a file can it handle?

There's no file size limit of ours. Files of several gigabytes work, because the file is read from your disk as it's needed rather than loaded all at once. What sets the pace is your device: a long 4K video takes longer on an older laptop than on a new one.

Which browsers does it work in?

Current versions of Chrome, Edge and Safari, and Firefox on a computer: any browser with the WebCodecs API for encoding video and sound. Some files, like iPhone HEVC video, need Chrome, Edge or Safari to decode. A computer is much faster than a phone for long videos.

More free tools for your videos

Silence remover

Cut the pauses and dead air out of any video or recording, automatically. Full quality, no watermark.

Video to text

Transcribe a video or a recording in 99 languages, with timestamps and SRT or VTT captions. Nothing uploaded.

Add captions to video

Auto-captions burned into your video, word by word, in five looks, clear of TikTok, Reels and Shorts buttons. Nothing uploaded.

Filler word remover

Find every um and uh in a video or recording, check them in the transcript, and cut them out. Free, no watermark.

Video compressor

Shrink any video to a target size or quality as an MP4 that plays everywhere. No upload.

Video resizer

Change any video to 9:16, 1:1, 4:5 or 16:9 with a blurred fill or a crop. Full quality, no watermark.

Extract audio from video

Save any video's sound as MP3, WAV or M4A, the original copied untouched where possible.

Video trimmer

Cut the start and end off any video, frame-exact or lossless with no re-encoding.

Loudness normalizer

Bring any audio or video to −14, −16 or −23 LUFS, peaks held under −1 dBTP. Video untouched.

Apply LUT to video

Grade a video with your .cube LUT or a free look, see it live, and download the MP4. Nothing uploaded.

TikTok safe zone checker

See TikTok's, Reels' and Shorts' buttons and captions over your video, and download 1080×1920 PNG templates.

Auto reframe video

Landscape to vertical with AI face tracking: the crop follows the speaker, smoothly, shot by shot. Nothing uploaded.

Blur faces in video

Every face found and followed automatically, then blurred or pixelated. Keep the ones you choose. Nothing uploaded.

Audiogram generator

A clip of your episode as a video: cover art, a moving waveform and word-by-word captions, in four shapes.

MP3 to MP4 converter

Any MP3 as an MP4 with your image on it, the whole file, ready for YouTube. Fast, even for an hour-long episode.

Bleep out swear words

Find every swear word in a video or recording, check them in the transcript, and bleep them out. Free, no watermark.

Teleprompter

Read your script to camera: scrolling at your pace or your voice's, mirrored for glass, and recorded as an MP4.