日本語English

Audio Noise Removal

To reduce background noise in a recording or meeting, choose the audio or video, set the strength and click “Remove noise”. Compare the result with the original, then save it as 16-bit WAV. Your file is never uploaded.

On-device・No uploadLimited on mobileWorks offline

Your files and input are processed only inside your browser on this device and are never uploaded to our servers.

Drop a recording, meeting or video

or

Choose one audio or video file (M4A, WAV, MP3, MP4, WebM and so on; whatever this browser can read). Length limits depend on the device: 20 minutes on a PC, 5 minutes on a phone.Accepts: MP4 / M4V / MOV / WEBM / MKV / M4A / AAC / WAV / OGG / OPUS / FLAC / MP3Paste (Ctrl+V) also works

How to use

  1. meeting.mp4audio is extracted

    Add audio or a video

  2. meeting (audio)hum reduced

    Reduce noise with AI (pick the strength)

  3. meeting_clean.wav16-bit WAV

    Compare, then save as WAV

Reduce background noise in a recording, compare before and after, then save as WAV

  1. Drop an audio or video file, or click “Choose file”. If there are several audio tracks, choose the one to use.

  2. Set “Strength of noise reduction”. Try the default first, and lower it if the voice starts to sound thin.

  3. Click “Remove noise”. Longer recordings take longer to process.

  4. Switch between “Original” and “Processed” to compare, check the waveform, then click “Save as WAV”.

What it does

AI noise suppression (RNNoise)

Separates speech from noise with the RNNoise neural model and suppresses the noise. The model is served from this site; nothing is sent elsewhere.

Adjustable strength

From 0% to 100%. The strength is the share of the removed noise that is taken away, so you can avoid over-processing.

Compare before and after

Switch between the original and the processed sound, with the waveforms side by side.

Processed on your device

The file is not sent anywhere. Processing runs in a Web Worker and can be stopped part way.

Works on the sound of a video

For MP4, WebM and other video, only the audio is processed and saved as 16-bit WAV.

Use cases

  • Make a recorded online meeting easier to hear by reducing air-conditioner or room hiss
  • Reduce background noise in a lecture or interview recorded on a phone
  • Take the audio out of a video, reduce its noise and save it as WAV
  • Clean up audio before transcription so that speech is recognised more easily

Supported formats

Input

Audio: M4A, AAC, WAV, MP3, OGG, Opus, FLAC and so on (what the browser can read)Video: MP4, MOV, WebM, MKV and so on (only the audio is used)

Output

WAV (16-bit PCM, same sample rate and channel count as the input)

Capability matrix

FeatureSupportNotes
Reduces steady noise (hiss, air-conditioner or fan noise)SupportedThe strength sets how much is removed. It does not remove every trace
Reduces short noises such as keyboard clicksPartialReduced, but a click that happens at the same time as speech is not fully removed
Fully separates a voice from noise where they overlapNot supportedWhere they overlap, the voice may also become quieter
Extracts vocals from music (vocal separation)Not supportedThis tool reduces noise; it does not separate music
Removes echo or reverberationNot supportedReverberation is not reduced much
Before and after comparison and waveform viewSupported—
Saving the picture of a videoNot supportedNo picture is kept; only the audio is saved as WAV
Saving as M4A, MP3 or OpusNot supportedThis version writes 16-bit WAV only
Loudness normalisation (LUFS)Not supported—
DRM-protected (encrypted) audio or videoNot supportedCannot be read

Limits & notes

  • Up to 2 GB and 20 minutes per file on a PC. Phones: 300 MB and 5 minutes.
  • The whole recording is held in memory while it is processed, so cut a long recording into parts first.
  • The output is 16-bit WAV at the input sample rate and channel count. One hour of 48 kHz stereo is about 690 MB.

Privacy & safety

Reading, noise removal and saving all happen in your browser. Audio and video are never sent to a server.

FAQ

How do I remove noise from a recording?
Choose the audio, set “Strength of noise reduction” and click “Remove noise”. Compare the processed sound with the original, then save it as WAV if it sounds right.
Does the voice get quieter too?
It can. Where noise and speech overlap, both become a little quieter. A lower strength keeps the result closer to the original.
Is my file uploaded?
No. Noise removal runs entirely in your browser, and the audio is not sent to a server.
Can I pull a voice out of music?
No. This tool reduces noise. It does not separate vocals from music.
Does it remove echo?
Echo and reverberation are not reduced much. The tool is mainly for background noise.
Can I save as MP3?
Not in this version: the output is 16-bit WAV. To change the format, convert the file with the audio converter afterwards.

Volume Normalization and LUFS

When sound is quiet, or the volume differs from file to file, set it to a target in LUFS. -14 suits YouTube and Spotify, -16 Apple Podcasts, and -23 broadcast. The tool measures first, then changes only the gain, and keeps the true peak under -1 dBTP. Your file is never uploaded.

No upload

Audio Converter and Video Audio Extractor

To get just the sound from a video, choose the video and an output format (MP3, M4A, WAV, Opus or FLAC). If the source audio is AAC, M4A is written without re-encoding, so the quality does not change. Your file is never uploaded.

No upload

Microphone Test

Start the mic to see input level (dBFS), waveform and spectrum. Clipping, noise and mute are detected automatically, and a guided speak test rates whether your mic is good enough for meetings. Nothing is uploaded.

No upload

Audio Editor

To cut unwanted parts of audio, drag on the waveform to select a range and press “Remove selected ranges”. To join clips, use “Join and edit” to set the order and any silence between them. You can save as WAV or MP3, or as M4A or Opus where your browser can write them. Your file is never uploaded.

No upload

Music Tag Studio

Edit and save the tags of MP3 and FLAC files in one table. Title, artist, album, track number, year, genre, lyrics and album art are supported, and garbled Shift_JIS tags common in old Japanese MP3s can be re-read and saved as UTF-8 / UTF-16. Audio data is not changed by a single byte, and everything stays on your device.

No upload

Screen Recorder

Record a screen, window or tab, your webcam and microphone without installing anything. Add a camera overlay or mix mic and tab audio. Recordings stay inside this device and are never uploaded.

No upload

Last updated: