Using whisper dictation on MacOS

[doctorguile]

I know dictation is not a function provided by org-ai but it was prominently featured in the demo video and @rksm also provided his config which greatly helps with figuring out what value to set for whisper--ffmpeg-input-device

https://gist.github.com/rksm/04be012be07671cd5e1dc6ec5b077e34

I'm using Ventura 13.6 and the dictation function failed silently, it turns out that ffmpeg is getting killed with "abort trap: 6"
because MacOS won't allow access to the mic unless the app requested it and user grant it explicitly.

And since emacs doens't do that by default we need to do something like this

https://superuser.com/questions/1441270/apps-dont-show-up-in-camera-and-microphone-privacy-settings-in-macbook

sqlite3 ~/Library/Application\ Support/com.apple.TCC/TCC.db

insert into access values  ('kTCCServiceMicrophone','org.gnu.Emacs', 0, 2, 2, 1, null, null, null, 'UNUSED', null, 0, 1698279008);

select * from access where service = 'kTCCServiceMicrophone' ;

.quit

I thought it might help if this gets mentioned in this repo somewhere or add as comments in the above gist

Cheers.

[rksm]

Great, thank you. There are steps in the readme for running tccutil.py from inside Emacs which should solve the same: https://github.com/rksm/org-ai/#macos-specific-steps. But I added a ref to this issue as well.