Your audio stays local
Decoding, mixing, noise gating and loudness metering all run in Web Audio. Not a single audio byte reaches a server — use it for interviews, internal meetings and unreleased course material.
Cut audio in your browser — it never leaves your device
A multi-track timeline, split and fades, a noise gate, 3-band EQ and compression, and loudness normalisation to -14 / -16 / -23 LUFS before export. Everything a podcast or a course recording needs, on one page. All processing happens on your device; nothing is uploaded.
No upload · No queue · Free for now
Hear changes instantly
The processing chain is parameters, not a rendered file. Change the high-pass, the gate or the compressor and the sound follows immediately. Preview and export share the same code path.
Loudness, by the standard
Integrated loudness is measured per ITU-R BS.1770-4 and can be normalised to -14 (streaming), -16 (podcast) or -23 LUFS (broadcast), always keeping at least 1 dB of peak headroom.
Close the tab, keep the work
The project and its sources autosave on your machine. Come back tomorrow and keep cutting. Drafts live in the browser too — never uploaded.
Tools deliver a file; a Studio lets you go back
Tools answer “give me a piece of audio”: paste text for a voice-over, remove one noise, submit, wait, download. A Studio answers “I need something I can still change”: it stores tracks, clip ranges and processing parameters, so tomorrow you can shorten the fade on the third segment by two seconds.
The same action feels different in each mode. Noise reduction in Tools is upload, wait, retry if you don't like it. In a Studio it is pull the threshold, listen for two seconds, pull a little more. The side that lets you change your mind is the side worth polishing on.
What this version does
Up to 8 tracks. Clips can be dragged, trimmed at both ends, split at the playhead, or cut automatically at pauses; each clip has its own gain and fades. Every track carries a processing chain: high-pass (cuts 60–140 Hz rumble), 3-band EQ (200 Hz / 1.6 kHz / 6 kHz), compressor (threshold and ratio) and a noise gate (strength and threshold).
Export 16 or 24-bit WAV at 44.1 or 48 kHz, downmix to mono, and normalise loudness first if you want. You can record straight onto the timeline (microphone permission required) or drop in mp3 / wav / m4a / ogg / flac.
What this version does not do
No AI voice-over, no voice cloning, no transcription. All three need model weights that can't run in a browser tab (the smallest usable TTS is still 82M+ parameters). We won't paper over that with an “AI” button that does nothing.
Also missing: MP3 export, time-stretching without pitch change, vocal/instrumental separation, audio extraction from video, and collaboration. For post-production, hand the WAV to your editing software — that's exactly why lossless export is the right default.
Three steps to a finished recording
- 01Import or record
Drop files onto the timeline, or hit record. Clips land at the playhead and can be moved anytime.
- 02Clean it up
Select the track, high-pass the rumble, add a little noise gate. Lift quiet material with the compressor, then check loudness.
- 03Export
Pick sample rate and bit depth, set a loudness target, render locally and download the WAV.
How it differs from an online audio converter
| OpenFilmo Audio Studio | Typical online audio tools | |
|---|---|---|
| Does audio get uploaded | No — everything runs in the browser | Uploaded first, then queued |
| Can you revise it | Project + 80-step undo, come back anytime | One file at a time; redo from scratch |
| Noise reduction | Noise gate with adjustable threshold and strength | Mostly fixed presets |
| Loudness | BS.1770-4 LUFS metering and normalisation | Rarely available |
| Cost | Free for now, not per file | Usually charged per file or per minute |
Frequently asked
- Is my audio uploaded?
- No. Decoding, mixing, gating and loudness metering all happen in Web Audio inside your browser; the server only delivers the page. Once the editor is open you can keep working offline.
- Which formats are supported?
- Import mp3, wav, m4a (aac), ogg and flac, as far as your browser can decode them (current Chrome, Edge and Safari all can). Export is WAV only, 16 or 24 bit.
- Why no MP3 export?
- There is no dependable MP3 encoder available in the browser: third-party libraries either carry licensing questions or are too heavy. WAV is lossless — converting it with your editing software or a transcoder is the safer route.
- Is the noise reduction AI?
- No. This version uses a noise gate: it measures energy to decide what counts as floor noise and attenuates everything below the threshold, with a fast-attack/slow-release envelope to avoid pumping. It works on steady hiss and pauses; it does not remove reverb or loud background music. Model-based denoising needs a GPU node, which we don't have yet.
- Will loudness normalisation clip?
- Normalisation computes gain from integrated loudness, then checks peaks: if hitting the target would exceed -1 dBFS, it stops at the clipping limit and the dialog tells you the target wasn't fully reached.
- Any size or length limits?
- A single source is capped at 30 minutes (longer files tend to blow up browser decoding) and 8 tracks. Drafts are stored in your browser; clearing site data clears them, so export anything important.