Online Voice Recorder

Record in your browser and save as MP3 or WAV

This online voice recorder records the sound from your computer's or phone's microphone in the browser, with nothing to install, and lets you download it as an MP3 or WAV file when you are done. While you record, a big clock and a live waveform of the last 10 seconds move on screen. If you need a break, pause and then continue in the same take. Each take you stop is added to the list below, and one tab can hold up to 60 minutes in total.

After you stop, you can choose the part to keep with the two handles on the full waveform, or use "Trim silence" to remove the empty time right after you pressed record and just before you stopped. The save panel shows how large the file will be for the format and quality you pick. The sound is not sent to a server. It stays in the memory of this device's browser, so download the takes you need before you close the tab.

When an online voice recorder helps

• Recording on a computer where you cannot install programs: on a work or school computer with installs blocked, a browser is enough to make a recording. The downloaded MP3 or WAV opens in ordinary players and editors.
• Practicing a talk or an interview answer: record the same script several times and the takes pile up in the list, so you can listen back to back and compare your pace and filler words.
• Keeping a record of singing or instrument practice: with "Noise reduction" off, the browser does not process the sound, so the recording stays closer to the original. Saved as WAV, it is easy to level or cut later in an editor.
• Sending a voice message by chat or email: saved as MP3 at 64 kbps, a minute takes about 480 KB, which is easier to fit under attachment limits. The estimated size is shown before you save.
• Trying out a video voice-over: record each sentence separately and trim the silence at the start and end, and it is easier to line the clips up with the picture in an editor.

How to use it

1. Press "Start recording" and allow the microphone. Once you allow it, recording starts at once, the status changes to "Recording" and the clock counts up. If you only want to check the device and the level first, press "Turn mic on" in the Input panel.
2. Speak or play. If the level bar at the top right turns the accent color, the sound was too loud and got clipped, so move a little farther from the microphone. Press "Pause" to stop for a moment and "Resume" to continue.
3. Press "Stop recording". The take goes into the list and its full waveform appears. "Listen" plays the part you keep, and clicking the waveform plays from that point. If you stop while listening, the next "Listen" continues from where you stopped.
4. Choose the part to keep. Drag the handles at the two ends of the waveform, or select a handle and move it with the arrow keys by 0.1 s (1 s with Shift). "Trim silence" cuts automatically around where the sound starts and ends.
5. Pick a file name and format, then download. For MP3, choose 64, 128 or 192 kbps. WAV is saved without compression. A downloaded take is marked as saved on its card in the list.

MP3 or WAV: which to save

MP3 is a lossy format that makes files smaller by dropping sound the ear barely notices. WAV usually holds uncompressed PCM samples as they are. The WAV from this recorder is 16-bit mono PCM. The MP3 is MPEG-1 Layer III, made at whichever of 32, 44.1 or 48 kHz matches your microphone's sample rate.

The table below gives the estimated size by recording length. MP3 is compressed at a constant bitrate, so a quiet recording of the same length has the same size. If your microphone runs at 44.1 kHz, WAV is about 8 % smaller than in this table and MP3 is almost the same.

How to choose Use 64 kbps for a spoken note you are sending, 128 kbps for everyday use, and 192 kbps for singing or instruments with a lot of high sound. If you plan to edit the recording, download WAV and convert to MP3 after editing, so it goes through lossy compression only once.

Tips for a clean recording

• Keep about a hand's width between the microphone and your mouth. Too close, and the air from "p" and "b" sounds hits the microphone with a thump. Too far, and the room echo grows. Placing the microphone slightly off to the side of your mouth reduces the puffs of air.
• Set your level with the meter. While you speak, the bar should move between about half and three quarters. If the bar turns the accent color, the waveform has been clipped, and lowering the volume later does not bring the lost sound back.
• Turn noise reduction on or off to suit the job. For speech only, such as a meeting or a voice memo, it lowers steady background noise like fans and air conditioners, and some echo. It is tuned for speech, though, so singing or instruments may sound different, and auto gain keeps changing the volume. Turn it off when you record music.
• Reduce room echo. A room with curtains, bedding or bookshelves echoes less than one with bare walls and a bare floor. Stay away from steady sounds such as a refrigerator or a computer fan.
• Use the device's own microphone rather than Bluetooth earbuds. Earbud microphones use the hands-free profile (HFP), which is mono, and its wideband speech mode is 16 kHz, so quality is lower than a built-in or wired microphone.
• On a phone, keep this tab in front the whole time. The page asks the screen to stay on while you record, but if you switch to another app the browser may pause audio. Nothing is recorded while it is paused, and a notice appears on screen.

Recording other people

Rules on recording conversations differ by country and region, and this tool cannot tell you whether a particular recording is allowed. Check the rules that apply where you are before you record other people.

• Tell people first: if you record a meeting, a class or a call, say so beforehand and follow the rules set by the organizer.
• Do not leave a recording running unattended: if you leave the room with the recorder on, it records conversations you are not part of.
• Keep recordings of other people private: share a file only with people who know it exists.

How recording and saving work

The browser's AudioWorklet (low-latency audio processing that runs on a separate thread) hands over the microphone sound in blocks of 2,048 samples (about 43 ms at 48 kHz) without gaps. Pause and stop take effect at these block boundaries, so the last 43 ms or so before you press Stop may not be in the recording. The recorder converts it to 16-bit integers and keeps it in memory in one-second chunks. If the microphone sends two channels, the average of the two is recorded as mono. The waveform is drawn from the loudest level kept for every 10 ms, so even a 60-minute take can show its full waveform at once.

• WAV: a 44-byte header (RIFF, fmt, data) followed by the 16-bit PCM samples as they are. Saving only the trimmed part needs no recalculation, since just those samples are cut out and joined, so the download is immediate.
• MP3: wasm-media-encoders, which is the LAME encoder compiled to WebAssembly, runs in a background worker and compresses the take 10 seconds at a time at the bitrate you chose. Progress shows on the button, and you can keep using the page.
• Estimated size: for WAV it is 44 + 2 × the number of samples, in bytes. For MP3, one frame holds 1,152 samples, so the size is (frames + 1) × 144,000 × bitrate (kbps) ÷ sample rate, in bytes. The real file is almost exactly this.
• Trim silence: the recorder looks at each 10 ms bin and finds the first and last place where the loudest sound is at least 10 % of the loudest value among sounds lasting more than 50 ms (very quiet takes use a fixed minimum) for more than 50 ms. It keeps 0.15 s before and 0.25 s after. A short spike like the click of the record button is not treated as sound and is also left out when the threshold is set, so a loud click does not make quiet speech count as silence.

Common mistakes

• Closing the tab without downloading: recordings are not kept on a server, so closing or reloading the tab removes them. If a take has not been downloaded, desktop browsers usually ask before you leave, but phone browsers may close without asking.
• Expecting the computer's own sound to be recorded: YouTube audio or the voice of the other person in a video call is recorded only as far as it reaches the microphone. With earphones, almost nothing is picked up.
• Missing the permission prompt: if you press "Start recording" and the status stays on "Connecting mic", look for the permission prompt near the address bar. If you blocked it once, allow the microphone in the site settings.
• Another app is using the microphone: if a video call or recording app holds the microphone, it may not open or no sound comes in. Close that app and press the button again.
• Recording with the wrong microphone: a webcam, monitor or headset microphone is sometimes picked by default. Turn the mic on first, choose the device in the Input panel and check the level meter before you record.
• Closing the tab during MP3 conversion: a long take takes a while to convert. Wait until the progress on the button finishes and the file downloads.

Limits

• It records one microphone as mono. Two-channel stereo and recording several microphones at once are not supported.
• One tab records up to 60 minutes in total, and a 48 kHz microphone uses about 346 MB of memory for 60 minutes. On a phone with little memory, it is safer to stop and download from time to time during a long recording.
• Editing is limited to trimming the start and end. Removing a section in the middle or joining several takes needs an audio editor.
• Recordings are not uploaded, so there is no share link. To send one, attach the downloaded file yourself.
• The noise reduction setting is a request to the browser and may not be applied as set on every device. After you turn the mic on, the Input panel's device info shows what was actually applied.
• Old browsers without AudioWorklet cannot record.

As of The source documents were checked on 2026-10-28.

Sources

• 01MDN: AudioWorkletA Web Audio feature for running low-latency audio processing scripts on a separate thread
• 02MDN: MediaTrackConstraintsSettings for requesting auto gain (autoGainControl), echo cancellation (echoCancellation), noise suppression (noiseSuppression) and channel count when opening a microphone
• 03MDN: Screen Wake Lock APIKeeps the screen from dimming or locking; released when the document is no longer visible
• 04Wikipedia: WAVThe most common WAV holds uncompressed linear PCM; RIFF format from IBM and Microsoft; under 4 GiB because of the 32-bit size field
• 05Wikipedia: MP3MPEG-1 Layer III bitrates of 32 to 320 kbit/s and sample rates of 32, 44.1 and 48 kHz, 1,152 samples per frame, encoder LAME
• 06wasm-media-encoders (GitHub)A library that runs the LAME MP3 encoder as WebAssembly (MIT license; LAME is LGPL)
• 07Wikipedia: List of Bluetooth profilesThe Hands-Free Profile (HFP) uses a mono voice channel; wideband speech (mSBC) in version 1.6 is a 16 kHz mono configuration

Frequently asked questions

Tools to use with it