Clean audio files with the browser tool
Use the tool to reduce steady hiss, mains hum, or room echo from one or more local audio files.
- Drop files into the tool or click browse files.
- Enable Remove mains hum, Reduce steady hiss/noise floor, and/or Reduce room echo/reverb tail. If mains hum is enabled, choose 60 Hz (US) or 50 Hz (EU/UK).
- Set Strength; the current file updates automatically.
- Compare Original and Cleaned, review the warnings, and adjust the controls if needed.
- Click Download cleaned file. Each result is a new 16-bit PCM WAV file with a
-cleaned.wavsuffix.
Choose the cleaning controls
- Remove mains hum uses the selected 60 Hz (US) or 50 Hz (EU/UK) frequency and four harmonics.
- Reduce steady hiss/noise floor uses spectral-subtraction processing.
- Reduce room echo/reverb tail uses spectral-gate processing and is off by default.
- Strength changes the aggressiveness of hiss reduction and room-echo gating. It does not change the fixed mains-hum notch settings.
Set the noise profile
Noise profile: uses the quietest 0.4 seconds automatically. The selector is disabled, so do not try to switch it directly.
- Drag across a noise-only part of the waveform.
- Click Use waveform selection to set that region as the manual profile.
- Click Clear (use auto) to return to automatic profiling.
Click the waveform to seek. A manual region appears as soon as you select it; the automatic region is shown after processing.
Process one file or a batch
For several files, add them to the queue; each queued file updates automatically with the current filter settings. Batch processing uses the current filter settings, but noise selections are stored per file. A manual selection for one file is not applied to every queued file; files without their own selection use their own automatic profile.
If the browser cannot decode a file, it is skipped during intake and the error message above the workspace identifies the file so you can choose another source.
Review and download the result
Use the Original/Cleaned playback buttons for A/B listening. The spectrogram has separate Original and Cleaned tabs, and you can click the spectrogram to seek. Batch processing provides a Download all (ZIP) download. Every cleaned file is newly encoded as 16-bit PCM WAV.
Limitations and privacy
This tool reduces steady, repeating noise with fixed filters. It does not perform stem separation or voice separation, remove arbitrary foreground sounds, or provide studio-grade restoration. The local implementation does not establish a browser-by-browser compatibility list, maximum file size, maximum duration, or guaranteed support for every format shown in the interface.
Files are read and processed with browser APIs and are not uploaded by the tool. Your original file is not modified.
Frequently asked questions
How does automatic noise profiling work?
The tool uses the quietest 0.4 seconds as the automatic noise sample. The automatic option is represented by the disabled Noise profile: selector.
How do I set a manual noise profile?
Drag across the waveform, then click Use waveform selection. Click Clear (use auto) to return to automatic profiling.
Does one manual profile apply to every queued file?
No. A manual selection is stored per file. Other files use their own automatic profile unless they have their own selection.
What format is the cleaned download?
It is a new 16-bit PCM WAV file with a -cleaned.wav suffix. A batch download is provided as a ZIP.
What happens if a file cannot be decoded?
It is skipped during intake and the decode error message identifies the file so you can choose another source.
Does the tool upload audio files?
No. Files are read and processed with browser APIs and are not uploaded by the tool.
Can it remove voices, stems, or arbitrary foreground sounds?
No. It does not perform stem separation or voice separation, and it cannot remove an arbitrary foreground sound.