I want to convert text to audio and back to text for error-checking.
I know how to get 'say' to go to a file but not a pipe. Apparently a third-party 'espeak' can go to a pipe.
But I don't find a man page or any shell commands for Dictation.
Another question pointed me to a company offering an app that can do it by GUI, but I'd rather do it by script.
I have reasons for wanting to do it by script, but the biggest is that the performance of Dictation is so crappy. The quality of recognition is still as good as ever, but where it used to transcribe on the fly, now it apparently buffers text and does the transcription only if the audio pauses. And the buffers apparently are so small (on a MacBook with eight GB RAM and swap on an SSD!) that the pauses have to be very frequent. Also, you have to tell Dictation all the punctuation. So a script could break the text at punctuation marks, and "say" the punctuation after each phrase.
The second reason for preferring a script is that the loopback app is expensive.
