How to record voice online without software in a private client-side recorder
When you click Start recording, this online voice recorder captures audio from your selected microphone directly into tab memory and encodes chunks locally every 1,000 ms. No audio frames, filenames, or metadata ever leave your device, and recordings vanish automatically when you close the tab.
Recording in 1,000 ms MediaRecorder chunks inside tab memory
Under the MDN MediaRecorder API documentation, the recorder calls mediaRecorder.start(1000) to flush encoded audio chunks into an in-memory array once per second. Pressing Pause suspends chunk capture and records a timestamp so paused intervals are subtracted accurately from the elapsed timer.
When you press Stop, all chunks assemble into a single browser Blob and appear under Your recordings with an HTML5 audio player. Before starting a long interview or narration session, you can check microphone levels and clipping before recording to confirm your input device is active and properly gained.
Reading live RMS volume from -60 dB to 0 dB
Alongside recording, the tool routes your microphone stream into a Web Audio AudioContext connected to an AnalyserNode with an fftSize of 2,048 samples. On every animation frame, it computes root-mean-square (RMS) amplitude across the buffer and maps decibels from -60 dB to 0 dB onto the horizontal Microphone level bar.
- Normal speech range: Clear speaking volume typically keeps the dark meter bar moving between 40% and 80% without hitting the top edge.
- Clipping warning: Any level exceeding 92% turns the meter bar red so you know to back away from the capsule or lower OS input gain.
Voice recorder M4A vs WAV comparison for sharing and editing
Every finished take in Your recordings offers two download buttons: the browser’s compressed native container and an uncompressed WAV export. Picking the right button depends on whether you care more about compact sharing or universal editing compatibility.
How browser codec selection picks M4A, WebM, or OGG
At startup, the tool queries MediaRecorder.isTypeSupported() against a prioritized list of audio MIME types. It prefers MP4 with AAC (audio/mp4;codecs=mp4a.40.2, saved as .m4a) where supported because Apple devices and messaging apps open M4A natively.
On browsers that do not encode MP4 audio, such as most desktop Linux or certain Windows Chromium builds, it falls back to audio/webm;codecs=opus (.webm) or audio/ogg;codecs=opus (.ogg). If you record several takes for a project or client review, you can bundle multiple saved audio takes into one ZIP archive after downloading them.
Converting takes to 16-bit PCM WAV off the main thread
Clicking Save .wav decodes the compressed take via AudioContext.decodeAudioData(), copies each channel’s Float32Array PCM buffer, and transfers ownership to a dedicated Web Worker. Inside the worker, encodeWav() writes a standard 44 bytes RIFF header followed by interleaved 16-bit signed integer PCM samples.
According to the Library of Congress WAVE Linear PCM format analysis, uncompressed Linear PCM inside a RIFF WAVE wrapper is the standard master format for audio editors and long-term preservation. Exporting to WAV avoids generation loss when importing takes into Audacity, Premiere, or transcription pipelines.
| Output option | Container and codec | Browser support | Best use case |
|---|---|---|---|
| Save .m4a | MP4 container, AAC (mp4a.40.2) |
Safari, iOS, macOS, newer Chrome | Voice notes shared with phones and Apple devices |
| Save .webm | WebM container, Opus codec | Chrome, Edge, Firefox, Android | Compact web playback and browser-based notes |
| Save .ogg | OGG container, Opus codec | Firefox and fallback engines | Lightweight archival where Opus is preferred |
| Save .wav | 44-byte RIFF header, 16-bit PCM | Every desktop and web audio app | DAW editing, podcasting, and lossless trimming |
Microphone speech processing vs music recording
The Reduce noise and echo checkbox in the options panel looks minor, yet it completely changes how your browser’s audio driver shapes incoming sound. Knowing when to leave it enabled and when to uncheck it prevents thin, pumping recordings.
What Reduce noise and echo changes in getUserMedia
Under the MDN MediaDevices getUserMedia specification, audio input requests accept three boolean DSP constraints: echoCancellation, noiseSuppression, and autoGainControl. When Reduce noise and echo is checked (the default), all three constraints are set to true.
That speech filter suppresses laptop fans, room reverb, and speaker feedback during spoken memos. However, unchecking Reduce noise and echo sets all three flags to false, passing raw, unprocessed audio straight from your audio interface or condenser mic so acoustic guitar, singing, or ambient field sounds retain their natural dynamics.
Switching microphones and pausing mid-take
If you plug in a USB headset or external XLR interface, select it from the Microphone dropdown before pressing Start recording. Both the Microphone selector and the Reduce noise and echo checkbox lock while recording or paused so the active MediaStream track never tears mid-take.
Each take can be renamed inline in the Your recordings list before saving, with invalid filename characters sanitized automatically. If you need to capture both a screen walkthrough and voiceover together instead of audio alone, switch to the browser screen recorder with audio.
Troubleshooting blocked microphones and clipping meters
Browser security rules strictly gate hardware access behind HTTPS and explicit user permission. Most recording failures trace back to denied site permissions, another video-call app holding exclusive device access, or an overloaded input gain knob.
When getUserMedia throws a NotAllowedError or SecurityError, click the padlock or camera icon in your browser’s address bar, switch Microphone from Block to Allow, and click Start recording again. A NotReadableError or AbortError usually means Zoom, Teams, Discord, or another open tab has locked the audio driver.
Under the MDN Web Audio AnalyserNode reference, time-domain analysis reads raw float samples between -1.0 and +1.0 without altering the underlying recording stream. If the Microphone level meter repeatedly flashes red above 92%, lower your microphone gain in system sound settings because digital clipping above 0 dBFS cannot be repaired after capture.
Tip: If you are setting up a new device or testing browser camera and scanner permissions in the same session, you can also run Scan QR code online locally without installing extra software.