Voice Recorder

Record and review one take

1. Select Record and respond to your browser's microphone permission prompt. Recording starts only for the current request. Cancel prevents a pending request from starting a take, although it cannot close the browser's prompt.
2. Use Pause, Resume and Stop. Pausing leaves the microphone open while audio is not added to the take. Hiding the page requests a stop; leaving the document discards unsaved data.
3. After finalization and initial playback readiness, review the recording and use Download recording. Its extension follows the actual native format. Discard recording removes the take before another can start.

Timing, native formats and final bytes

Active estimate adds confirmed recording intervals and excludes pauses and finalization. Session estimate includes pauses from the confirmed start through the stop request. Neither measures speech or the exact duration encoded in the file.

The tool tries advertised WebM/Opus, Ogg/Opus and MP4 audio paths, then checks the native type hints and file family. The resulting download is WebM, Ogg or M4A, not a universal MP3. A reported codec is distinguished from an unreported one.

Stop requests release the tool's microphone tracks promptly, while final data events are collected in order before one result is assembled. A fatal error, oversized take or failed readiness check discards the take instead of exporting a partial result.

The native player may report a duration or leave it unknown. Media-reported duration is shown separately from the estimates. Initial playback readiness is not full-file decoding or a guarantee that every other player can open the file.

Pause, timing and format examples

If recording starts at one second, pauses at six, resumes at ten and Stop is requested at thirteen, the active estimate is eight seconds and the session estimate is twelve. A later final stop event does not add drain time to those estimates.

A WebM take can be ready for playback while its native duration is not reported. The display then keeps Media-reported duration as Not reported; it does not substitute the active estimate.

A take stopped when the page became hidden keeps that reason if finalization succeeds. A canceled permission request that later succeeds has its tracks stopped and creates no recording. Neither case starts again automatically.

Capture and playback limits

Targets are ten minutes of active estimate, fifteen minutes of session estimate and 32 MiB of retained output. Event delays, sleep and native buffering can delay stop processing or allocate resources before JavaScript acts; these are not exact cutoff or process-memory guarantees.

A secure context, microphone permission and the required native recording controls and formats are needed. Device, permission, encoding and playback failures remain possible. Browser microphone processing, silence, device quality and another app's capture are outside an audible-speech guarantee.

Pause does not release the microphone. Stop, cancellation and discard stop this tool's tracks. Hidden pages request a stop and pause playback, while page departure discards unsaved data. There is no background, locked-screen or crash-recovery promise.

This tool has no trimming, mixing, transcription, uploaded audio, saved history or MP3 conversion. Native formats and seeking support vary by browser and player. A download click starts browser handling; it is not proof that a file was saved.

Explicit microphone capture without tool uploads

The tool adds no recording upload, remote conversion, saved take, input-bearing link or telemetry containing audio. Permission is requested only after Record. Existing site services, browser and operating-system behavior and extensions are outside a universal privacy promise.