Meeting transcription uses speech recognition to convert audio into written dialogue. A system may identify speakers, attach timestamps, recognize domain terms, and connect each statement with the source recording.
Transcripts can support notes and action extraction, but they may mishear names, numbers, accents, or overlapping speech. Participants need appropriate notice and consent, while important decisions should be checked against the recording or confirmed by attendees.
ELI5
Meeting transcription turns what people say in a meeting into written text. The transcript can make the discussion searchable and easier to review without replaying the entire recording.
For example, someone can search the transcript for a project name and jump to the moment it was discussed. The system may mishear a deadline or person's name, so important details should be checked and participants should know when recording occurs.
What can a meeting transcript include?
It can include spoken text, speaker labels, timestamps, links to audio, corrections, notes, decisions, and extracted tasks.
Why must meeting transcripts be reviewed?
Speech recognition can mishear names, numbers, technical terms, accents, and overlapping speech, changing the meaning of important details.

