Every other tool here sends your file to a server when you export, because encoding to MP3 or FLAC requires an encoder. The recorder does not. It captures from your microphone through the browser's MediaRecorder API, assembles the result in memory, and writes a WAV file straight to your downloads folder. No format conversion is needed, so there is nothing to send and nothing to delete afterwards. The audio never leaves the machine it was recorded on.
Your browser will ask for microphone permission the first time. That prompt is the browser's, not ours, and the permission applies only to this page. Recording stops the moment you press stop or close the tab.
Getting a usable recording
- Distance matters more than any setting. A hand's width from the microphone is roughly right; at arm's length the room becomes louder than you are.
- Soft furnishings beat bare rooms. A recording made on a bed or a sofa has noticeably less echo than the same words in a kitchen.
- Wired beats wireless. Bluetooth headsets compress audio before it reaches the browser, and that compression cannot be undone later.
- Do a five-second test and listen back before recording something long. It costs nothing and catches the wrong input device, which is the most common problem.
WAV is uncompressed, so recordings are large — roughly 10 MB per minute in stereo. That is deliberate: it keeps the original at full quality. Run it through the Audio Converter afterwards if you need something smaller.
What to do with the recording
Because the output is WAV, it is a lossless starting point for anything else. Trim the dead air off either end with the MP3 Cutter, remove steady background noise such as a fan or air conditioning with the Noise Remover, even out the level with the Audio Compressor if you moved around while talking, then convert to MP3 with the Audio Converter for sharing.
There is no time limit on a recording, but very long sessions are held in memory until you stop, so a phone with little free memory may struggle past an hour. If you are recording something long, splitting it into shorter takes is safer than relying on one continuous capture.