Remove Background Noise from Audio

AI removes hiss, fan hum, air-con and white noise from your recordings. No install, no upload.

Drop your audio or video file here or click to select

MP3·WAV·M4A·FLAC·OGG·WebM·MP4 and more · video files: audio track is extracted · processed entirely on your device

Choose file

0 B uploaded — this page has not sent your files to any server?

Stereo file detected — noise removal is processed in mono (channels are averaged).

Original

Denoised

Private recordings stay private — AI noise removal runs right inside your browser. Your file is never uploaded anywhere, and there is no upload wait.How this differs from upload-based sites — most converter sites upload your video or recording to a server, process it there, and promise to delete it after a while. Until that happens, your file is sitting on someone else's server. vidmoa has no upload step at all. The more sensitive the file — a recorded meeting, a phone call, family footage — the more that difference matters.

Keep the voice, lose the hum: cleaning fan and air-conditioner noise out of a recording

Air handling in a meeting room, a laptop fan or the steady hiss of a recorder is easy to ignore while recording and hard to ignore when you listen back. This tool runs a speech-focused AI noise suppressor inside your browser, keeping voices while turning down the steady noise underneath. Drop in a video and only its audio track is processed; nothing is uploaded.

A closer look

The engine is RNNoise, an open-source model published by Xiph, adapted to run in the browser. The first time you press the button it downloads the roughly 120 KB engine; after that, your file is converted to 48 kHz mono, sliced into 10-millisecond frames, and each frame is judged for how much is speech and how much is noise before the noisy part is lowered. Before the real pass, the first second is run through once so the model can settle on the character of the room, and then processing starts again from the beginning. There is a single strength with no slider, so if the result sounds unnatural, the honest answer is to keep the original.

The output is always mono: stereo files are averaged into one channel, and a note appears when that happens. You can save a 48 kHz 16-bit WAV, about 5.5 MB per minute, or a 128 kbps MP3, about 940 KB per minute. A video goes in but only audio comes out, so putting the cleaned sound back means replacing the track in a video editor. The whole file is decoded into memory at once, so recordings several hours long are safer split into parts first, for example with the Audio cutter.

How to use it

Drag in an audio or video file, or press Choose file. One file is processed at a time; to swap it for another, use the remove button at the end of its row.
Press Remove noise. The status runs through “Decoding audio…”, “Resampling to 48 kHz mono…” and “Removing noise… 40%”, with an extra loading step the very first time.
When it is done, Original and Denoised players appear side by side. Starting one pauses the other, so you can flip between them on the same passage and check that word endings survived.
Press Save as WAV or Save as MP3. The file is named with the date and time, for example “denoised_20260927_1430.wav”.

When it helps

Lectures recorded over a fanClean up a class captured while the laptop fan was running so that, on replay, you hear the lecturer instead of the whir.

Interviews before transcriptionTidy an interview with office air handling underneath so it is easier to transcribe.

Narration for a videoPull the audio out of a narrated clip recorded at home, reduce the air-conditioner noise, and swap it back in while editing.

Phone voice memosClean a voice memo recorded next to a humming fridge or fan before forwarding it, and save it as MP3 so it stays small.

Common snags

It says it “Could not read audio from this file.”Either the browser cannot decode the codec or the file has no audio track. Try converting it to MP3 or WAV with the format converter first, and for a video, play it to make sure there is sound in it.

The voice sounds watery, or word endings are clipped.When the noise is as loud as the voice, or the noise itself is speech-like, such as a TV or other people talking, the model struggles to tell them apart. With no strength control, keep the original for files like this; moving the microphone closer next time helps more than any processing.

The result is quieter than the original.If a stereo file had sound on one channel only, such as a lavalier plugged into one side, averaging the channels roughly halves the level. Raise the volume in an editor after saving.

Room echo stays and background music sounds odd.The model separates voice from noise, so the reverb of a large room is treated as part of the voice and kept, while music can lose its instruments along with the noise. It is not recommended for videos with music underneath.

Tip If you leave a second or two of silence with just the room tone at the start of a recording, the warm-up on the first second has pure noise to learn from, which tends to make the opening sound more even.

FAQ

What kinds of noise does it remove?

It is strongest on steady background noise: fans, air conditioning, computer hum, white noise, tape hiss and keyboard clatter. The AI keeps the voice and suppresses everything else, so it works best on speech — voice memos, meetings, lectures, podcasts. Music is not recommended, since instruments may be suppressed too.

How is the audio processed?

Audio is converted to 48 kHz mono before the noise is removed. Stereo files are downmixed by averaging both channels. You can save the result as lossless WAV or as MP3.

Does it work on video files?

Yes. Drop an MP4, WebM or MOV and the audio track is extracted and denoised. The result is saved as an audio file (WAV/MP3).

Is my file uploaded?

No. AI noise removal runs inside your browser, so your file never leaves your device — and even long recordings start instantly.

Is it free?

Completely free. No limits, no watermark.