Edith,
My initial hope for this application was to do just what you describe. As far as I understand the Google API, however, you can't both simultaneously record audio and be presented with immediate visual feedback (words appearing as you speak). Someone
please correct me if I'm wrong-- the documentation offered by Google is somewhat sparse.
The original version of this software worked by first saving the audio on a server through NanoGong and then subsequently sending the entire chunk to the API for a transcription. The advantage of this approach was that the audio resided on the
server for optional review by the instructor or student. The file transfer and analysis, however, could take up to 30 seconds for feedback-- sometimes longer with groups working simultaneously. I also found it problematic to dial in the perfect bitrate and
filetype for Google's transcription engine. Transcriptions seem more accurate with the live approach, and ultimately our pilot group here at SLU ended up preferring the current implementation.
All that said, the kind of program you describe might be possible in the future and it remains an avenue I hope to explore.
Cheers,
Dan