Remove background noise
Constant background noise (a fan, air conditioning, computer hum, microphone hiss) makes recordings tiring to listen to. This tool learns the noise from the quietest moments of your recording and subtracts it throughout. We explain its limits plainly below.
Free · no sign-up · your file is processed in your browser and never sent to a server
No file handy? Try it with a 20-second Turkish meeting recording.
How to use
- Choose the recordingDrop your speech recording. For the best result, have a few seconds of noise only at the start or somewhere in between.
- Pick a strength and compareChoose light, medium or strong. Compare the first 10 seconds with the "Before" and "After" buttons.
- Apply and downloadProcess the whole recording with the setting you like, listen and download.
How it works
- FilterRumble below 80 Hz (air conditioning, traffic, mic handling) is cut; optionally hiss above 9 kHz too.
- Spectral subtractionA frequency profile of the noise is taken from the quietest 10% of the recording and subtracted from the sound at every moment.
- Noise gateWhen no one is speaking, the remaining noise is turned down within milliseconds, and it opens again when speech starts.
What it does well, and what it can't do
| Type of noise | Result |
|---|---|
| Fan, air conditioning, computer hum | Good: steady noise drops noticeably. |
| Microphone hiss, tape noise | Good; even better with the high-cut filter. |
| Mains hum (50/60 Hz) | Partly: the fundamental is cut, upper harmonics may remain. |
| Music, TV or crowd in the background | Weak: changing noise can't be separated from speech. |
| Echoey room | Can't: reverb isn't noise but reflections of the voice itself. |
This is a classic signal-processing method that runs in your browser without any server, so don't expect studio-grade cleanup. On the strong setting, voices may sound muffled or "underwater". That's why you should compare first; for most recordings "medium" is the best balance.
For cleaner recordings
- Bring the microphone closer to your mouth; your voice rises relative to the noise.
- Stay silent for 2–3 seconds before you start speaking; the tool learns the noise more accurately from that part.
- Switch off air conditioning, fans and notification sounds; record in a room with soft surfaces.
Privacy: your recording never leaves your device
Everything happens in your browser. Your file is not sent to Voiser's servers or stored anywhere, and it's cleared from memory when you close the tab. That's also why the tool stays fast on a slow connection.
Common questions
Is the noise removed completely?
Steady noise drops noticeably, but often doesn't disappear entirely. Changing noise (music, crowds) and reverb can't be cleaned with this method.
Which strength should I choose?
Try "Medium" first and compare the first 10 seconds. If the voice sounds muffled, switch to "Light"; if the noise is still obvious, try "Strong".
Can I use it on music?
It's designed for speech. On music, subtraction can remove fine instrumental detail too; use the light setting with the high-cut filter off.
Is my recording sent to a server?
No. All processing happens in your browser; your recording never leaves your device.
Can I transcribe the cleaned recording?
Yes. A cleaner recording means a more accurate transcript; you can upload the result to Transcription.
Related tools
- Volume boosterMake quiet recordings louder, or normalize them, without clipping.
- Silence removerAutomatically find and remove long silences in a recording.
- MP3 cutter and audio trimmerSelect any part of an audio file on the waveform and cut it out.
- Extract audio from videoSave the audio track of your own videos as MP3, M4A or WAV.
- Audio to text converterTurn a recording into a timestamped transcript with speakers.
- Online voice recorderRecord audio in your browser and download as WAV, MP3 or WebM.
Cleaner audio, more accurate text.
Transcription turns speech into time-stamped text with separate speakers. Start with a free account.