Back to Free Tools
FREE TOOLS
Audio Noise Reducer
Local stationary-noise reduction with a user-selected noise profile.
Supported local audio
WAV, JSON
Browser only
Runs locally in your browser. Your files, pasted content and generated output are not uploaded. Reset clears this tool’s working data.
This tool needs JavaScript. If it does not load, refresh the page and try again.
How to use this tool
- Choose one local recording and listen for a part containing only representative background noise.
- Enter its start and end seconds (0.25–30 seconds), choose reduction strength and smoothing, then process locally.
- Listen to the reduced audio and removed component for wanted-signal damage. Inspect the report, adjust settings if needed, and download.
Examples
Profile one second of fan noise before speech, then reduce the same steady background noise through the recording.
Limitations
- One 128 MiB input, five minutes, mono/stereo and 32 million decoded samples including decoder priming; browser codec support varies.
- A noise-only profile 0.25–30 seconds long must be inside the recording. RMS below -100 dBFS is too quiet to use. Speech or music in the profile can damage wanted audio.
- Designed for steady noise. Variable traffic, clicks, competing voices and reverberation are not reliably removed; no intelligibility or universal denoising guarantee.
- Spectral reduction can introduce watery/tonal artifacts. Maximum reduction is a per-frequency ceiling, not a promised overall reduction. Stereo masks are linked.
- Output 48 kHz, 16-bit WAV retains the original timeline and channel count; tags are omitted. Any global attenuation needed to prevent clipping is reported.
- Processing is browser-local in one cancellable worker with a 120-second deadline. Decoded buffers can require substantial memory on long stereo inputs.
Related tools
No items found.