- add init_logging option to allow users to use their own logging framework
- reduce realtime transcription load by only transcribing if new frames arrive
- wrong arguments for self.stream.read
- make sure shutdown() wakes up the recording_thread
- clear text storage and audio frames on audio queue clear
- add webrtc that does not need visual studio to build
- remove unneeded import