Prepare the meeting file
Use an MP3, WAV, M4A, or MP4 recording up to 200 MB or 20 minutes. A close microphone and limited room noise make review easier.
A meeting transcript is most valuable when it helps someone find a decision, action item, or exact quote later. This workflow turns a meeting recording into timestamped draft text that can be checked against the source before it becomes a set of notes.
Prepare a private meeting draft with participant labels, timestamps, and terminology checks.
Before processing, add participant names, project names, product terms, and acronyms to the names and terms field. This gives the local model useful context and creates a review signal for terms that need attention. It does not automatically identify who is speaking.
After transcription, assign Speaker 1 through Speaker 4 to segments during review. This is a manual labeling workflow, designed to make a conversation easier to read without promising automatic diarization in a noisy room or a recording with overlapping speech.
Use an MP3, WAV, M4A, or MP4 recording up to 200 MB or 20 minutes. A close microphone and limited room noise make review easier.
List participant names, teams, products, acronyms, and decisions that should be checked after transcription.
Use timestamps to replay a decision, assign manual speaker labels, edit action items, and export a text or structured version of the meeting.
Timestamped segments let a reviewer return to the original audio when an action item, number, or decision is unclear.
Speaker labels improve readability after review. They are not automatic identification, and overlapping speakers still need human judgment.
Names and project vocabulary can be supplied before processing and checked afterward, reducing the chance that a critical term is overlooked.
Example output only. The actual transcript depends on the recording, language, microphone, and background noise.
[00:12:44.880] Speaker 2: Alex will share the revised launch checklist by Thursday.
No. The current local build lets you assign Speaker 1 through Speaker 4 manually to transcript segments.
Yes. Add names, teams, products, and acronyms to the names and terms field so they can guide the model and appear in review suggestions.
The initial browser limit is 20 minutes or 200 MB per file. Longer recordings need to be split or handled by a future cloud workflow.
Treat it as a searchable draft. Decisions, dates, owners, and commitments should be verified against the recording before distribution.