Transcribe a meeting with speaker labels
Meeting Notes transcribes English speech and separates voices so you can name the speakers afterward. You can open a recording or record a meeting in the tool; processing runs on your device.
Open Meeting NotesUse clear audio
Keep the original recording and follow the recording permissions expected by participants. Distant voices, room noise and overlapping speech make transcription and speaker separation harder.
Transcribe the meeting
Open the recording or use the recording flow. The first use downloads speech and speaker models. Longer recordings take more time and memory than a short sample.
Name the voices
Listen to clear passages from each detected voice before assigning names. Similar voices and brief interjections can be misclassified. The model separates voices; it does not verify a person’s identity.
Check and download the notes
Review names, dates, numbers, responsibilities and negations before copying or downloading the text. Leave uncertain statements uncertain rather than converting a guess into an action item.
Fixing speaker groups
During the current audio session, adjust the speaker count when needed and reassign a passage that belongs to another speaker. Giving groups the same speaker name merges them, so use distinct labels for different people who share a first name. Compare clear passages before merging: short interruptions can sound different from a long turn. After a reload, text and names remain but audio and voiceprints do not, so automatic regrouping is unavailable.
Turn a checked transcript into useful minutes
The transcript is evidence to review, not an automatic record of agreement. After listening to important passages, separate decisions from suggestions and write action items with an owner and due date only when those were actually stated. A compact note might use the structure below; it is an example you organize from the recording, not a promise of an automatically generated summary.
Decision: what was agreed
Action: task — owner — due date, if stated
Open question: what still needs an answerChoose meeting notes or live dictation
Meeting Notes is useful when a recording contains several voices and speaker grouping matters. Whisper Dictation is simpler for capturing your own spoken draft. Neither eliminates the need to check names, dates and negations. For a noisy long recording, test a representative section first so you can assess the model’s output before investing in the entire transcription.