Skip to main content

Meeting recording & transcription — Regression Test

Prerequisites​

  • Microphone and System Audio permissions granted
  • A working microphone (speak into it during the test)

Steps​

  1. On the Meeting tab, open a meeting note (from Up Next) or click New Note.
    • Note detail view opens on the My Notes tab with a Record Meeting button
  2. Click Record Meeting and speak clearly for 10–15 seconds.
    • Recording starts and a Pause Recording button appears
    • The audio waveform responds to your voice
    • Live transcription bubbles appear in the Transcript tab as you speak
    • The microphone selector (current input device name) is visible in the Transcript tab
    • When you pause speaking for about a second between sentences, completed text appears without waiting for a long silence; network delay can affect the visible timing
  3. Speak a final sentence, then click Pause Recording.
    • While the pause request is pending, the control shows Pausing... with a loading indicator; repeated clicks do not start another pause request (PR #1511)
    • Recording pauses and a Resume Recording button appears
    • The final sentence is retained in the transcript after the pause finishes (PR #1315)
    • The Summarize button is available
  4. Click Summarize and wait.
    • "Generating your summary…" shows while it processes
    • A formatted summary appears in the Summary tab
    • On desktop, Summarize ends the paused meeting: the banner and tray no longer say the recording is paused (PR #1317)
  5. Open the Transcript tab and click Copy transcript.
    • A "Transcript copied to clipboard" confirmation appears
    • Pasting elsewhere shows the transcript with speaker labels (You / Speaker)

Variations​

  1. Click Resume Recording, speak a few more seconds, then Pause Recording. On desktop, step 4 ended the meeting, so Resume Recording starts a new recording on the same note; it stays available within the grace period after the last transcript.
    • Resume Recording is shown after the summary and starts recording on the same note
    • New transcription is appended to the existing transcript
    • Summarize reappears (new transcript since last summary); clicking it produces an updated summary covering all content
  2. On the Transcript tab, open the microphone selector and switch to a different input device mid-session (resume first if paused).
    • The device list shows available inputs with the active one checked
    • Recording continues after switching
  3. Click Pause Recording, then navigate away to another screen and back to the note.
    • All transcription and the summary are still present
    • The recording state (paused) is correctly restored

Variation — recording from Apple Watch (new)​

Deep, TCC-dependent coverage of this flow lives in the repository at apps/clients/docs/testing/manual-tests/watch-recording-tests.md, which is outside this Docusaurus site. This is the one-pass regression variation.

  1. On a paired Apple Watch with the app's watch companion installed, open it and start a recording from New note or from an upcoming meeting under Up Next.
  2. Speak for 10–15 seconds, then stop the recording from the watch.
  3. Open the iPhone app (or desktop/web on the same account) and find the note.
  • The note appears with a transcript of what you said, the same as a note recorded on desktop or mobile
  • No duplicate note is created if you look at the note again after the upload finishes

Variation — speaker names in a group call (new)​

Steps 1–3 cover names guessed from the calendar event's attendee list, so start those recordings from an Up Next event whose attendees have names — a New Note has no calendar candidates. Hold those calls in Google Meet in a browser without the HabitatZero for Meetings extension: on macOS desktop, a name read from Zoom, Microsoft Teams, or the extension replaces a calendar guess, so step 1's Best guess label would not appear. Step 4 covers those meeting-app names.

  1. From Up Next, record a calendar meeting with two or more other named attendees (not a 1:1). Have the others introduce themselves or address each other by name, and let enough of the call transcribe that more than one non-"You" speaker appears.
    • At least one speaker other than you shows an inferred name with a small Best guess label next to it — not just "Speaker 1" / "Speaker 2"
  2. Hover (or tap, on touch) a speaker's name in the transcript and click the pencil icon (Correct speaker name) next to it, then type a different name and press Enter (or tap away to blur).
    • The name switches to an editable field, pre-filled and selected
    • After committing, every bubble from that same speaker updates to the new name, not just the one you edited
    • The Best guess label is gone from the corrected name
    • Copy transcript, then paste into a text editor: the corrected name appears beside that speaker’s lines, rather than a generic Speaker label
  3. From Up Next, record a 1:1 calendar meeting (just you and one other attendee whose name is on the event).
    • The other participant's side of the transcript shows that attendee's name rather than a generic "Speaker" label
  4. macOS desktop only. Make sure HabitatZero has Accessibility access (System Settings → Privacy & Security → Accessibility). Join a call with two or more other people in the Zoom desktop app or the new Microsoft Teams app, open any note (a New Note is fine — these names come from the meeting app, not the calendar), and click Record Meeting. Have each other person talk on their own for a few seconds.
    • Each other person's bubbles show the name the meeting app shows for them, with no Best guess label
    • Your own bubbles still read You
    • In Zoom, the participants list opens by itself if it was closed — this is expected

Variation — transcription stopping partway (new in 5.0.0)​

  1. Start a recording and let a little speech transcribe.
  2. Turn Wi-Fi off for about ten seconds, keeping the recording running.
  3. Turn Wi-Fi back on and speak again.
  4. Stop the recording.
  • While offline the app tells you transcription stopped — it does not carry on as though it were still transcribing
  • The recording itself keeps running; you are not silently dropped out of the meeting
  • After stopping, the meeting is saved and listed with whatever transcript was captured
  • Note whether speech after reconnecting was transcribed — either answer is useful, but say which you saw

Tray Now-meeting actions (PR #1139, #1145, #1317)​

On packaged desktop with recording enabled, use an ongoing disposable calendar meeting.

  1. Open the system tray menu and start recording from the Now meeting's action.
    • The corresponding meeting note opens and recording starts once
  2. Pause, then resume from that meeting's tray actions.
    • The note's recording state agrees with the tray; pause/resume does not require switching to the main window
  3. Open the note from the tray, then end it with Summarize in the note.
    • The same note opens, and after ending it the tray no longer presents it as a paused recording

For platform speaker naming and post-call voice-group checks, also run the repo's apps/clients/docs/testing/manual-tests/meeting-speaker-attribution-tests.md.