Closed DamienDeepgram closed 3 months ago
Reduce the amount of audio being transcribed that does not contain speech in the friend apps.
Use a Voice Activity Detector (VAD) in the apps to decide what to transcribe.
Today
With a VAD
If you apply this logic to 12 hours per day of recording most likely only a small percentage of the day contains speech
To be implemented in the backend
Backend sounds good will close this one
Goal
Reduce the amount of audio being transcribed that does not contain speech in the friend apps.
Approach
Use a Voice Activity Detector (VAD) in the apps to decide what to transcribe.
Example
Today
With a VAD
If you apply this logic to 12 hours per day of recording most likely only a small percentage of the day contains speech
Benefits