How to Remove Background Noise from Audio
The simplest way to remove background noise from audio is to drop the file into Vocal Cut's free Noise Reducer, pick Voice or Music mode, press the process button, set how much reduction you want, and export. It takes a handful of clicks, it's free with no account, and it runs in your browser, so your audio never leaves your device.
The button-pressing is easy. Whether the result sounds natural comes down to two decisions you make first: what kind of noise you have, and which mode suits the recording.
First, identify the noise you're dealing with
"Background noise" covers several different problems, and software handles some of them far better than others. Before you process anything, listen to a few seconds where nobody is talking and decide which of these you hear.
- Steady, broadband noise. Hiss from a cheap mic or preamp, a laptop fan, air conditioning, the low wash of room tone. It sounds roughly the same from start to finish. This is the easiest kind to reduce, because the software can learn what it sounds like and subtract it.
- Tonal noise. Electrical hum or buzz from mains power, at 50 Hz or 60 Hz depending on your country, plus its harmonics. It's a pitched drone rather than a hiss. Because it sits at predictable frequencies, a notch filter can take it out cleanly.
- Intermittent noise. Traffic passing, a dog barking, keyboard clicks, a door closing. These come and go and don't look like the rest of the noise, so a steady-noise filter mostly misses them. An AI speech model does better, but a loud bark under a sentence is still hard to remove without damage.
- Reverb and echo. The sound of the room bouncing back off hard walls. Strictly speaking this isn't noise at all; it's a smeared copy of the voice itself. Noise reducers help only a little here, and pushing them hard tends to make speech sound hollow.
If you want the longer explanation of why these behave so differently, How AI Noise Reduction Works goes into it. For this guide, the short version is: steady noise and hum are very fixable, intermittent noise is partly fixable, and echo is mostly a recording problem.
Voice or Music mode: pick the right engine
The Noise Reducer has two separate engines behind its Voice and Music buttons. They work in completely different ways, and choosing the wrong one is the most common reason a result sounds bad.
| Voice mode | Music mode | |
|---|---|---|
| How it works | AI speech model (GTCRN) | Spectral processing, not AI |
| Best for | Podcasts, voice memos, interviews, narration | Songs, full mixes, instrument recordings |
| What it removes | Everything that isn't a voice | Steady hiss, hum and broadband noise |
| Download | Small AI model (about 0.5 MB) on first run, then cached | None |
| Output | Mono, processed in the speech band (16 kHz) | Full-band, stereo preserved |
| Strength | Strong | Deliberately conservative (about 16 dB of reduction at most) |
Voice mode is the closest thing on the site to an AI audio enhancer for speech. A small neural network listens for the human voice and suppresses everything else, which is why it copes with intermittent sounds that a simple filter can't. The catch is that it removes everything that isn't a voice, and the tool shows a warning about it: "Voice mode is a speech model — don't run songs through it." Put a song through it and the instruments, pads, cymbals and reverb tails get eaten along with the noise.
Music mode builds a noise profile from the file automatically, so you don't need to select a noise-only section, and then gently pulls that steady noise down across the full frequency range. It aims to reduce noise, not erase it, so the music itself stays intact. It also has a checkbox, Also remove 50/60 Hz electrical hum (auto-detected), which works out whether your hum is at 50 or 60 Hz and notches the fundamental and its harmonics up to the fifth.
A rule of thumb: if the recording is mostly someone talking, use Voice. If it's music, or a voice recording where you need to keep music underneath, use Music.
What you'll need
- An audio file in MP3, WAV, FLAC, M4A, AAC, OGG, Opus, or WebM audio.
- A file no larger than 100 MB and no longer than 12 minutes. The tool holds the audio and its analysis in memory, so it caps length honestly. For a longer recording, split it into parts with the Audio Cutter first.
- A modern browser, ideally on a laptop or desktop.
- Headphones. Leftover noise and processing artifacts are much easier to hear on headphones than on laptop speakers.
There's no account to create and nothing to pay.
How to remove background noise from audio, step by step
- Open the Noise Reducer and drop your file onto the box labelled "Drop or upload audio to reduce noise," or click it to browse. Once it loads, you'll see the file name, its length, its sample rate and whether it's mono or stereo.
- Choose a mode. Click Voice for speech or Music for songs and mixes, using the table above as your guide.
- Turn on hum removal if you need it (Music mode). If you heard a low electrical drone, tick Also remove 50/60 Hz electrical hum (auto-detected).
- Press the process button. It reads Reduce noise (Voice mode) or Reduce noise (Music mode). A progress bar shows a real percentage, the current stage (for example "Downloading AI model" on the first Voice run, then "Reducing noise"; or "Analyzing noise profile" and "Rebuilding audio" in Music mode) and an estimate of the time left. Processing runs in a background worker, so a long file can take a while; let it finish.
- Preview the result. Press play. You'll see a slider running from Original to Denoised, with a readout such as "Noise reduction: 75%". It starts at 75%.
- Set the mix. This slider is the most important control in the tool. Drag it toward Original to blend some of the untouched recording back in; drag it to 0% to hear the original on its own for comparison. Changes apply live while it plays, and the export uses whatever mix you settle on, so there's no need to process again. If the voice sounds slightly watery or thin at 75%, try somewhere between 50% and 65%.
- Export. Pick a format in the selector: WAV (the default, lossless), MP3, or Opus where your browser supports it (Opus doesn't appear on Safari). Then click the export button, for example Export WAV. Use Reset File when you want to start over with a different recording.
How to remove background noise from a video
The Noise Reducer accepts audio files only, so cleaning up a video's sound is a three-step job:
- Extract the soundtrack with Video to Audio. How to Extract Audio from a Video has the full walkthrough.
- Clean it in the Noise Reducer following the steps above. Talking-head clips, tutorials and vlogs usually suit Voice mode; a clip with background music you want to keep needs Music mode.
- Put it back. Export as WAV, import it into your video editor, mute or delete the original audio track, and line the cleaned file up with the start of the clip.
Sync is rarely a problem, because noise reduction keeps the file's duration: if the cleaned audio starts where the original started, the words will land on the lips. One thing to know is that Voice mode outputs mono. For narration and dialogue that's normal, since a single voice is effectively mono anyway, and your editor will play it in both speakers. How to Clean Up Audio for YouTube Videos covers the rest of the creator workflow.
Getting a cleaner result
The voice sounds watery, robotic, or underwater. That's over-reduction: the tool is removing parts of the voice along with the noise. Lower the mix slider. Setting it to around 50–65% usually brings the natural sound back while keeping most of the improvement.
The music got eaten. You probably used Voice mode on a song or a recording with music behind the speech. Click Reset File, load it again, and use Music mode instead.
There's still noise in Music mode. That's by design. Music mode caps its reduction at around 16 dB so it can't hollow out a mix, and it only targets steady noise. If the leftover is a drone, try the hum checkbox. If it's hiss in a particular range, a gentle cut with the Equalizer can help; How to EQ a Podcast Voice for Clear Speech explains where to cut and boost.
Clicks, bumps or a cough are still there. Short, one-off sounds are the hardest kind of noise. If they fall between words, the simplest fix is to cut them out with the Audio Cutter. If the gaps between sentences are where the noise lives, the Silence Remover can trim those pauses.
It sounds quieter than before. Removing noise removes energy, so the file can feel lower in level afterwards. Run the export through the Volume Normalizer to bring it back up to a consistent loudness.
Prevent noise at the source
Every bit of noise you avoid while recording is noise you don't have to remove later, and prevention never introduces artifacts. Getting the microphone closer to your mouth is the single biggest improvement, because your voice gets louder relative to the room. Mic Placement for Clean Voice Recordings covers distance and angle, and Record Clean Vocals at Home on an Audio Recorder covers choosing a quiet space and setting levels.
After cleanup: once the noise is under control, the Audio Enhancer can add EQ, compression and loudness in one pass. It's a signal-processing chain rather than an AI tool, and Audio Enhancer: How to Improve Audio Quality walks through what each stage does.
Frequently asked questions
Can I remove background noise from audio for free? Yes. Vocal Cut's Noise Reducer is free to use with no account or sign-up. It runs in your browser, so processing happens on your own computer or phone instead of a server. Voice mode uses a small AI speech model for podcasts and interviews, Music mode uses spectral processing for songs, and you can export the cleaned file as WAV, MP3, or Opus where your browser supports it.
Does noise reduction hurt voice quality? It can if you push it too far. Heavy reduction removes some of the voice along with the noise, which sounds watery, thin, or robotic. The Noise Reducer guards against this with a mix slider that blends the original back in. It starts at 75% reduction; lowering it to around 50–65% often sounds more natural while still removing most of the background noise.
Can I remove background noise from a song? Partly. Use Music mode, which reduces steady hiss, hum, and broadband noise across the full frequency range and keeps the stereo image. It's deliberately conservative, with a maximum of about 16 dB of reduction, so it lowers noise rather than erasing it. Don't use Voice mode on songs: it's a speech model and will strip out instruments and reverb along with the noise.
How do I remove background noise from a video? Extract the soundtrack with a video-to-audio tool, clean that audio file in a noise reducer, then replace the original track in your video editor with the cleaned version. Line it up with the start of the clip. Noise reduction keeps the file's length, so the words stay in sync with the picture. Vocal Cut's Noise Reducer accepts audio files only, not video files directly.
Is my audio uploaded anywhere? No. Vocal Cut's Noise Reducer processes your file inside your browser, and your audio never leaves your device. Downloads only go the other way: in Voice mode, the browser downloads a small AI model (about 0.5 MB) the first time you use it, then caches it. Music mode needs no model download at all.