Clear audio improves the chance of a useful transcript. Transcription and speaker labels can be wrong, so review the result before relying on it or sharing it.
Before recording
Tell participants you are recording and obtain the permission you need.
Move away from machinery and wind where practical.
Keep the microphone close to the speakers and avoid overlapping speech.
State important names, parcel references and agreed dates clearly.
On the web
Open a consult session and allow microphone access when starting its recording. Check the visible recording status and transcript display. Browser/device permissions and the connection affect recording and live transcription. Do not assume a particular browser guarantees successful recording or accurate diarisation. Stop using the session controls, then choose Complete Session when ready to finish and start summary processing. Stopping alone leaves it open if you choose Keep Working. Check saved status and processing before reviewing the summary.
In a completed session's Summary tab, Download Audio requests the saved recording. If preparation fails, use Retry where offered. Preparing audio can take time; a stopped recording does not guarantee that a downloadable file is ready.
On iOS
Start a recording from Consults or the farm's available consult action. Allow microphone access. Use the recording screen's pause/resume and stop controls. Native consult recording uploads audio chunks for later processing; it does not show the web's live transcript pane. After stopping, check the session detail for its processing state, summary and expandable Transcript.
Voice notes are separate
A vault or field-capture voice note is not the same as a consult session. On the web, use the offered voice-note control in Records. On iOS, choose voice/audio capture where available and check the upload queue. Transcription availability and timing depend on the recording type and processing result.
Speaker diarisation labels voices; it does not guarantee that the system knows each person's identity or learns persistent voice profiles from an introduction.
