Voice recorder

What this tool does

It records sound from a microphone, right in the page, using the browser's own MediaRecorder. You pick the microphone, choose whether the browser should clean up the sound (echo cancellation, noise suppression and automatic gain), press Record, and watch the level meter while you speak. You can pause and resume as often as you like; the pauses are left out of the file. When you stop, you can listen back, download the recording in the format your browser produced, or convert it to a WAV file. The audio never leaves your device: there is no upload and no account.

How to use it

1. Pick a microphone. Until you have allowed the microphone once, browsers hide the device names, so the list may only say "Default microphone"; after the first recording the real names appear.
2. Decide on the clean-up. For a voice memo or a call, leave all three on. For music, or when you already use a good microphone in a quiet room, turn them off: noise suppression can make an instrument sound watery, and automatic gain keeps pumping the volume up and down.
3. Type a file name if you want one, then press Record and allow the microphone when the browser asks.
4. Watch the level while you speak. It shows the level in dBFS and says in words whether it is too quiet, good or too loud. Aim for "good": peaks well below 0 dBFS but clearly above the noise.
5. Press Stop, play it back, and Download it or Download WAV.

Which file you get

Browsers do not all record the same format. This tool asks for Opus audio in a WebM file first (Chrome, Edge and Firefox), then Opus in Ogg, then AAC in an MP4 file (Safari), and names the file to match: .webm, .ogg or .m4a. It asks for 128 kbit/s, which is roughly 0.96 MB per minute (128,000 bits × 60 seconds ÷ 8); the browser may choose a different rate, and the size shown under the player is the real one.

WAV is uncompressed: the browser decodes your recording and the tool writes 16-bit PCM at the decoder's sample rate. At 48 kHz that is 96,000 bytes a second per channel, or about 5.76 MB a minute in mono, six times the compressed file. Convert to WAV when an editor, a transcription service or a form insists on it; it does not make the sound better, because what the compression removed is gone. Conversion holds the whole recording in memory, so it is limited to 30 minutes.

Tips for clean sound

• Put the microphone about a hand's width from your mouth, slightly to the side, so plosives (p, b) do not hit it straight on.
• Record in the smallest, softest room you have. Curtains, a sofa and a bookshelf absorb echo; bare walls and a tiled floor add it.
• Turn off fans and notifications, and close the laptop's other apps that might grab the microphone.
• Record ten seconds of silence first and look at the meter: if the room itself shows above about -50 dBFS, the noise will be audible in the file.

Long recordings stop by themselves after 120 minutes. To cut the start and end off afterwards, use the audio trimmer; to hear a script read aloud before you record it, try text to speech. There is a longer guide in how to record clean voice audio on a laptop.