Audio Noise Remover
Give voice recordings a cleaner background. Adjust the strength, compare before and after, and keep the version that sounds right.
Up to 50 MB · 3 minutes · mono or stereo. Supported formats depend on your browser.
WAV, MP3, M4A, OGG, FLAC, WEBM • up to 50 MB
Cleanup strength
Start around 70%. Increase for more background reduction, or lower it if the voice sounds too processed.
Made for speech
- Try voice notes, interviews, lessons and podcasts.
- Fan noise and hiss may be reduced. Results vary with the recording.
- Overlapping voices, echo, clipping and music can remain or sound different.
- Mono and stereo are preserved. WAV output can be larger than your input.
Private by default
Your recording stays in this browser. No account, subscription or API key is needed. Recent files storage is optional and applies only to the completed result.
Clean up a voice recording
1. Open your audio
Choose a short voice recording. Your browser checks and decodes it locally.
2. Find your balance
Apply cleanup, compare both versions, and adjust the strength if needed.
3. Keep the result
Download a WAV with the same decoded duration and channel count.
Frequently Asked Questions
How does the audio noise remover work?
RNNoise uses a neural speech model to reduce background noise. Processing runs locally in a WebAssembly worker, independently for each audio channel. Adjust the cleanup strength to blend the processed sound with your original recording.
Which recordings work best?
Try spoken voice notes, interviews and podcasts with fan noise, hiss or a steady background. Results depend on the recording. Strong cleanup can change the voice, and the model cannot reliably separate overlapping speakers, repair clipping or remove all echo. It is designed for speech, not music restoration.
Are my recordings uploaded?
No. Your file is decoded and processed in your browser. The RNNoise engine loads from this website. Saving a result to Recent files is optional and stores it on this device only.
Which formats and file sizes can I use?
Choose an audio file up to 50 MB and 3 minutes long, in mono or stereo. Common inputs include WAV, MP3, M4A / AAC, OGG, FLAC and WebM, depending on your browser’s decoders. Unsupported or protected files cannot be opened. Large recordings may exceed your device’s available memory.
What format is the cleaned download?
The output is a 16-bit PCM WAV file at 48 kHz, preserving the decoded recording’s channel count and number of audio frames. Compressed inputs are decoded and may be resampled to 48 kHz, so the WAV can be larger than the original. This tool reduces noise; it is not an audio compressor.
What does the strength slider change?
At 0%, the output keeps the decoded original sound. Higher values blend in more of the neural cleanup; 100% uses the full processed signal. Click Apply cleanup after changing the slider, then compare before and after at a comfortable listening volume.