The online voice recorder records a piece of sound straight from the browser microphone, without opening the ffmpeg workbench first. Once it starts, a live waveform and level meter are shown, and you can pause, resume and stop; while idle, the browser's “echo cancellation / noise suppression” can be toggled. Recording uses getUserMedia + MediaRecorder underneath and asks for WebM/Opus first.

After stopping you can download the WebM directly, or decode it locally with AudioContext and export WAV. There is also a “Convert to MP3 (send to workbench)” button: only when you choose that step does the recording reach the audio/video workbench as a local File, to be transcoded by ffmpeg.wasm.

1. 01Grant the microphoneon an HTTPS page click “Start recording” and allow the browser to use the microphone; if permission is refused the page suggests checking the settings.
2. 02Record and watchthe live waveform and the level bar at the bottom follow the sound; click “Pause” when you need to and “Resume” afterwards.
3. 03Stop the recording“Stop” produces a local result and shows its duration, file size and a player.
4. 04Choose the export“Download WebM” directly, or “Download WAV” to decode and convert locally; send it to the audio/video workbench when you need MP3.

Browser requirements

a modern browser with getUserMedia and MediaRecorder, reached over HTTPS or from a trusted local environment.

Encoding

recording prefers WebM + Opus, and the final capability depends on what the current browser's MediaRecorder supports; the page cannot record MP3 directly.

WAV export

it relies on the browser's AudioContext decoding the WebM it just recorded; when that fails, the page suggests downloading the WebM and converting it in the workbench.

Permissions and devices

recording is impossible when the browser refuses the microphone, the system disables the device or another app holds it.

What it does not do

there is no multi-track support, editing, denoise-strength control or automatic transcription; send the result to “Audio Cut and Merge” or “Audio Denoise” to edit it.

Microphone data is recorded inside the current browser by MediaRecorder and held in an in-memory Blob; it is never uploaded, and no ffmpeg CDN request is made. Only when you actively send the result to the workbench for MP3 conversion does the workbench download the ffmpeg core for the first time.