Processing logic needs to know if it has already identified that the person has started speaking so that it can check if person has stopped speaking. If this state is not known, entire energy buffer will need to be processed to identify if the user has both started speaking and stopped speaking and processing time will vary based on how long the buffer is.
If the state of user having started speaking is known, one can just look back for a fixed duration to identify if the user has stopped speaking. Processing time will be constant in this case.
Processing logic needs to know if it has already identified that the person has started speaking so that it can check if person has stopped speaking. If this state is not known, entire energy buffer will need to be processed to identify if the user has both started speaking and stopped speaking and processing time will vary based on how long the buffer is.
If the state of user having started speaking is known, one can just look back for a fixed duration to identify if the user has stopped speaking. Processing time will be constant in this case.