When I record a meeting, I cannot record the next back-to-back meeting because Hedy takes a long time to process the first one. I end up missing important information.
I would like to use the Large Model for a better quality transcript, and that understandably takes even longer to process. I also don't want my phone's battery to be drained, as I won't have time to charge it as I move from meeting to meeting.
Maybe processing can start when I plug into a charger, and pause when I go off charge? Maybe it can be scheduled for overnight?
Perhaps Hedy's biggest constraint is the processing power on the device. If processing can be deferred, additional opportunities may become available:
  • There is less pressure for (near) real-time results. Some features can be synchronous and others can be asynchronous.
  • It can work on slower devices.
  • Advanced features can be added, to be executed asynchronously.