▶TranscriptReel
Home / Tools / Voice Recorder

Browser voice recorder

Record straight from your microphone and save the take as a file. The level meter tells you before you start whether the input is loud enough to transcribe well. Nothing leaves the tab.

Microphone permission is required. Your browser will ask once. The audio is captured and held in this tab only — there is no server, no upload and no storage, so a reload discards the take.
0.0selapsed
—peak level
—sample rate
—channels
Input level. If this stays in the first fifth of the bar, the recording will be too quiet to transcribe reliably — speak closer or raise the input volume first.
Idle. Press Start recording.

Recording for transcription rather than for listening

A memo you record for a human ear and a memo you record for speech-to-text want different things. For a person, atmosphere and distance are fine. For a model, the goal is one voice, close to the microphone, with no competing sound, and a level that fills the meter but never touches the top.

Two failures cause most of the bad transcripts people blame on the tool. The first is a quiet take: if the peak level sits far below the top of the meter, the model is working with a fraction of the dynamic range it needs, and you cannot fix that afterwards by boosting the file — the boost raises the room noise along with the voice. The second is a long silence at the head of the take, because people press record and then get settled. Trim that before transcribing.

Both problems are visible in this page before you upload anything: the level meter shows the first, the elapsed counter shows the second.

Why does the browser ask for permission every time?

Because there is no stored consent for this origin in that browser profile, or because you dismissed the prompt instead of answering it. If the prompt never appears, the microphone may be blocked for the whole browser in your system privacy settings.

Is there a maximum length?

Nothing is enforced here, but the take lives in memory and is converted to WAV in one pass, so a very long recording can fail on a low-memory device. Ten minutes of stereo audio at 44.1 kHz is about 100 MB of raw samples.

Can I record a system sound instead of the microphone?

Not from this page. Browsers only give access to an input device, and capturing what the computer is playing needs a screen-share capture with audio, which is a different permission and is not supported the same way in every browser.

Does this page send my audio anywhere?

No. There is no fetch call, no upload and no storage. The recording exists only in this tab and is gone when you close it.