Free Online Audio Editor with AI Noise Removal

A free online audio editor with AI noise removal

MES Audio Editor is an Audacity-style editor that runs in your browser. Open audio or the sound of a video, cut and rearrange it on several tracks, fade in and out, set the loudness for YouTube or podcasts, and remove background noise with a neural network. Nothing is uploaded: decoding, editing, the AI models and the export all run on your device.

How the noise removal works

The default model is DeepFilterNet 3 (Hendrik Schröter and contributors, MIT / Apache-2.0), a speech-enhancement network that removes fans, air conditioning, traffic, hiss, keyboard and room noise while keeping the voice natural. It is an 8.6 MB download the first time, kept by your browser afterwards, and runs on the processor through WebAssembly at roughly 10 to 30 times real time on a laptop (a 10-minute recording takes well under a minute). The audio is processed at 48 kHz and converted back to your project's rate. RNNoise (Xiph, BSD) is a smaller, faster and gentler alternative. The Spectral gate needs no download: select a few seconds of noise only, press "Learn noise from selection", then apply it; it suits steady hiss on music too, where the speech models are not the right tool. Strength limits how far the noise is pushed down, so you can keep a little room sound. Hum removal notches 50 or 60 Hz mains hum and its harmonics.

Use Preview to hear the selection processed, then switch between A = original and B = processed while it plays (key B). Speech models are trained on voices: on music or singing they can remove instruments too, so preview first.

Questions

Which files can I open?

Anything your browser can decode: MP3, WAV, M4A / AAC, OGG / Opus, FLAC and the audio track of MP4, MOV and WebM videos (MKV and AIFF depend on the browser). Several files open as separate tracks you can line up.

How do I make my audio the right loudness for YouTube or a podcast?

Effects → Normalize / loudness → choose −14 LUFS (YouTube, Spotify) or −16 LUFS (podcasts, Apple) with a −1 dBTP true-peak ceiling. The meter shows the integrated loudness, loudness range and true peak before and after. The export dialog can also do it on the way out.

What does "Enhance speech" do?

One click for a voice recording: AI noise removal, a low cut for rumble, a voice-clarity EQ, a gentle compressor and loudness to −16 LUFS. Each step can be switched off.

Which formats can I export?

WAV (16-bit, 24-bit or 32-bit float), MP3 (64 to 320 kbps, encoded in the page with lamejs) and M4A / AAC where the browser has an AAC encoder (Chrome, Edge and Safari). Export the whole project or only the selection, and markers as a text file Audacity can import.

Is my work saved?

The project is saved in this browser (IndexedDB) a few seconds after each change, so you can restore it after closing the tab. Nothing is sent to a server.

Can I edit the sound of a video?

Yes: open the video, clean and edit its sound, export a WAV or M4A, and put it back with the Video Compressor or your video editor. To shrink audio files, the MES Compressor makes small MP3s.

Keyboard shortcuts?

Space play / pause, Ctrl+Z / Ctrl+Shift+Z undo / redo, Ctrl+X / C / V cut, copy, paste, Delete, Ctrl+I split, Ctrl+T trim, Ctrl+L silence, M marker, R record, + / − zoom, N noise removal, F fill the window. Press ? in the editor for the full list.