1
S1
pass
Start phone mic recording, inject TTS audio, verify Deepgram transcribes successfully
✓
2
S2
pass
Open capturing page and verify real transcript segments are visible with 'Listening' status
✓
3
S3
pass
Kill the backend process to break the WebSocket connection
4
S4
pass
Check home screen recording bar shows 'Recording, reconnecting' with snackbar 'Still recording — reconnecting to transcription...'
✓
5
S5
pass
Restart backend, wait for WS reconnect, verify status returns to 'Listening'
✓
6
S6
pass
Stop the recording and verify processing begins
