Here's my use case: I have scripts for talks that are 90-95% accurate to what the speaker will actually say. I'm looking for a service where I can upload those scripts, and as the speaker is speaking, a machine transcription service will automatically return results in real time—using both the scripts and the built-in recognizer to deliver higher quality results than the built-in recognizer alone.
I'm aware that with the Google Cloud Speech API, I can get streaming results, and with phrase hints I can supply some context on input, but the time limit on streaming sessions (1 minute) and the requirement to both chunk my input into limited phrases and limit the number of phrases overall are both deal breakers.
Any other ideas?
