VOICE CLEAN · AUDIO UTILITY
Reduce steady fans, hum and hiss, balance speech and export a clean WAV for podcasts, lectures or video. It works in your browser; your recording is not uploaded to ANTB.
Or drop a supported audio or video file anywhere
3 of 3 free recordings remain today.
Daily limit reached. Sign in or create a free account to clean more recordings.
Loading the demo…
Drag the slider to compare the noise. Click the picture to jump. Space plays and pauses, B switches between versions.
Fixed demonstration. The demo uses one preset so everyone hears the same comparison. Choose or record your own audio to unlock the controls and download.
Fans, air conditioning, hiss and room tone
Brings quiet and loud words closer together
Less boxy room sound, more presence
Adds body to thin laptop and phone mics
Softens sharp S and T sounds
| Measure | Unfiltered | Filtered |
|---|---|---|
| Voice above background higher is cleaner | – | – |
| Loudness | – | – |
| Peak level | – | – |
| Length | – | – |
Mono WAV, 48 kHz, 16-bit
Made for one voice recorded somewhere ordinary: a home office, a classroom, a car, a kitchen table.
Fans, air conditioning, computer whir and hiss. The noise is learned from the pauses in your own clip, then turned down throughout.
50 and 60 Hz mains hum, desk bumps and traffic rumble below the voice are filtered out before anything else happens.
Quiet and loud words are brought closer together, then the whole clip is set to a standard loudness so it plays at the right level everywhere.
Clarity removes the cardboard-box sound of small rooms and lifts speech presence. Warmth gives body back to laptop and phone mics.
The de-esser softens sharp sibilance only at the moments it happens, so the rest of the voice keeps its brightness.
Optionally trims silences longer than half a second, which tightens lectures, voice notes and first-take narration.
What it won’t do: separate two voices, remove music, catch short sudden noises like a door slam or a bark, or remove heavy echo from a large room. For those, the best fix is to re-record somewhere quieter.
Every step runs on your device, in this order, each time you change a setting.
Yes. Guests can clean three recordings each day. Sign in or create a free account to process more.
One successfully opened or recorded audio clip counts as one use. Changing settings, comparing the filtered and unfiltered versions, and downloading that result do not use another credit. The built-in demonstration never counts.
No. Audio processing happens inside your browser using the Web Audio API. Your recording is not sent to ANTB.
Common WAV, MP3, M4A, AAC, OGG, FLAC and WebM audio files are supported. Voice Clean can also read audio from some MP4 and MOV videos. Exact support depends on your browser. The limit is 15 minutes on computers and 5 minutes on phones.
Use −16 LUFS for podcasts and most voice work. Use −14 LUFS if the audio is going on YouTube or Spotify, −19 LUFS if you want more headroom, and −23 LUFS for broadcast delivery under EBU R128.
No. Voice Clean is designed for steady background sound such as fans, air conditioning, hiss and hum. It does not separate voices, remove music or catch short sudden noises, and it cannot remove strong room echo.
A single voice is naturally mono, and WAV keeps every detail with no extra compression, so the file is ready for podcast hosts, video editors and further editing.
It is designed for current mobile browsers that support the Web Audio API. Recording needs microphone permission. Phone processing is limited to five minutes to reduce memory-related crashes.
Voice Clean estimates background noise and loudness; results vary with the recording and browser. Listen to the enhanced version before using it. Only process recordings you have permission to use. The tool is provided as-is and is not a substitute for professional audio restoration.
What has changed in Voice Clean, newest first.
