Transcription
Recordings are automatically transcribed in our supported languages. If commonly used words or phrases such as names or industry jargon are not transcribed correctly, please add them to your custom word dictionary. (Settings > Custom word dictionary)
Summaries
Summaries are automatically generated using your preferred summarization language. (Settings > Summarization language)
Saved Voices (Speaker Labeling)
How to label voices
Navigate to Settings > Saved voices
Press "Show guided tour"
We will automatically label voices going forward with 20 or more seconds of cumulatively labeled audio.
Automatic voice labeling is not retroactive
Tips for labeling voices & troubleshooting
Only the last 90 seconds of manually labeled audio is used to automatically label voices. Voice labeling is not perfect and won't work 100% of the time, however, labeling "clean audio" will lead to a higher chance of a voice being identified. The audio is "cleanest" when:
The desired voice is within 6 feet of the Pendant's microphone
The Pendant's microphone is not obstructed (behind clothing)
Only the desired voice is speaking (not being talked over)
The desired voice is speaking at a normal volume
There is little to no background noise
If a voice sounds significantly different in different situations (i.e. on a Zoom call and in-person), you will likely see better results creating two different speakers. If your saved voice is never or rarely automatically labeled, this is likely because:
The voice has less then 20 seconds manually labeled (see progress at Settings > Saved voices)
Labeling more clean audio always helps!
The voice sounds different then the labeled audio (i.e. Zoom vs in-person)
Too much "unclean" audio has been labeled
If you suspect too much "unclean" audio has been labeled, it is recommended to delete the saved voice (Settings > Saved voices > Trash icon) and start fresh.
β