Why so many recordings end up as WebM
WebM is the format web browsers produce when a site records audio or video. Chrome, Edge and Firefox all record WebM through the standard recording API built into the browser. That means a large share of browser-based tools save WebM files: online voice recorders, screen recorders that run in a tab, and Chrome extensions that capture Google Meet or other video calls.
Some web apps let you download recordings as WebM as well. If a file came from something you used inside a browser rather than a desktop app, there is a good chance it is WebM.
If you have not recorded anything yet, you may not need a separate recorder at all. The upload box on this page can record from your microphone directly in the browser and send the recording for transcription when you stop.
What is inside a WebM file
WebM is based on the Matroska container. The audio is usually Opus, sometimes the older Vorbis, and the video, if there is any, is VP8, VP9 or AV1. Many browser voice recorders save audio-only WebM files with no picture at all.
Opus was designed for voice over the internet, so it keeps speech clear even at low bitrates. For transcription, a WebM recording is usually as good as the microphone and room it was recorded in. You do not need to convert WebM to MP3 or MP4 first. Upload it as it is and the audio is extracted on our side.
Common WebM problems and how to handle them
Browser recordings have a few quirks that are worth knowing about before you upload.
- Missing duration: WebM files written by browsers often have no length stored in the header, so players show 0:00 or cannot seek. These files still upload and transcribe normally.
- Missing audio in screen recordings: when Chrome asks what to share, tab audio is only included if the share audio option is on. A screen recording made without it may contain your microphone only, or no sound.
- One side of a call: some call recorders capture either the tab audio or the microphone, not both. Play the file and check that you can hear everyone before you upload.
- Cut-off recordings: closing the tab or the browser mid-recording can leave a file that ends early. The transcript will stop where the audio stops.
Accuracy for browser and call recordings
Recordings made through a laptop microphone are usually clear for the person sitting at the laptop. Accuracy drops for remote participants whose audio arrived through a weak connection, because dropouts and robotic-sounding stretches are hard for any speech recognition to read. People talking over each other, echo from speakers, and keyboard noise also reduce accuracy. Clean stretches of speech transcribe very well.
If you record calls often, wear a headset. It keeps the other participants' voices from leaking back into your microphone, which cuts echo and doubled words in the transcript.
The language is detected automatically from the audio, or you can choose it per file. Calls that switch between Chinese and English mid-sentence are supported.
Using the WebM transcript
For a recorded call or meeting, export Word or PDF notes, or use the AI prompt export to paste the transcript into ChatGPT or Claude with a request for a summary. On paid plans, speaker labels show who said what, and you can rename them to the participants' names. For a screen-recorded tutorial or demo, export SRT or VTT captions after correcting any terms in the editor. A read-only share link lets you send the transcript to someone without an account.
Recorded calls can run for hours. Uploads are sent in parts and resume after a dropped connection, and paid plans allow much longer files than the free plan. Your recordings are never used to train AI models, and you can delete them at any time.