WEBM is the format the web records in. Browser-based screen recorders, Chrome extensions for capturing a tab, Loom in some export settings, OBS when the output is set for browser playback, and the MediaRecorder API that many web apps use for voice notes all produce it. If you recorded something inside a browser window, there is a good chance the file ends in .webm.
The audio inside is Opus or, in older files, Vorbis. Both are efficient speech codecs, and Opus in particular keeps voices clear at low bitrates, so WEBM recordings generally transcribe well. The problems people run into are not about quality. They are about tools that do not know the container, and about a quirk of how browsers write these files, described below.
Transcrify accepts WEBM directly. The container is opened on our servers, the Opus or Vorbis stream is decoded and the words are transcribed in any of more than fifty languages, detected automatically.
Browsers write WEBM files as a stream and often never go back to fill in the total length. The file plays fine, but any program that asks "how long is this?" gets no answer, and a progress bar in a video player jumps around for the same reason. Before uploading, this page reads the duration of your file in the browser to check it against the free limit. If the duration is missing, the upload stops with a message that it cannot read the duration.
The fix takes a minute. Open the file in any editor or converter that can re-save it, for example HandBrake, VLC with the convert function, or a quick export from a desktop video editor, and save it again as WEBM or MP4. The re-saved file has its length written in the header and uploads without complaint. Loom and most meeting platforms produce files with correct headers, so this mostly affects raw recordings from small browser tools.
If you record regularly with a browser extension and hit this every time, check whether the extension offers an MP4 output or a "finalize recording" option. Either one avoids the extra step.
A large share of WEBM files are narrated screen recordings: a walkthrough for a colleague, a reproduction of a bug, a short product demo. The transcript turns that narration into text with a timestamp on each segment, which gives you a written version for documentation, a ticket or a knowledge base without watching the recording again. Search the text, find the moment, and the timestamp tells you where in the video it happens.
Meeting recordings downloaded as WEBM behave the same way, with the addition of speaker labels. Each person the model can distinguish gets a consistent label through the file, which is enough to follow who said what in a call with a handful of participants.
Screen recordings often show internal tools and unreleased work, and the narration explains them. The speech models run on hardware we operate, and the audio is not forwarded to a third-party speech API. Guest transcriptions are kept under the same rules and move into your account if you later create one.
As a guest you can transcribe 10 recordings of up to 3 hours each and read or copy the text. A free account adds the editor, your transcription history and downloads in SRT, VTT, ASS and TXT, with 5 hours across 10 files. AI chat with the recording, summaries and export to PDF or Word are part of the paid plans.