Audio woes with AI
I recently got a M4 Mac mini, more on that later but along with that, I also got a Focusrite Scarlett Solo 4th gen, that has been working quite well up until now.
Yesterday I decided to try out an app called SuperWhisper, as I hoped that it would help speed up my workflow, in particular, writing these blog posts for example.
I installed it via my setapp subscription (referral link if you’re interested, we both get a free month), launched it and tried to dictate something. It didn’t see the mic, which was weird, so I checked the system preferences - security settings, all looked good there.
I decided to try ChatGPT and see if that worked, same problem. didn’t see the mic.
Ok, lets try something else, launched Quicktime, made an audio recording no problem. Launched Logic Pro, made a quick voice recording with no problems. Tried OBS, made a quick screen recording and voice over, no problems.
Went back to ChatGPT - tried again, still not working.
Checked System preferences - security (and accessibility) - ChatGPT (and SuperWhisper) were both allowed.
Checked System preferences - sound - Mic is being picked up, the Scarlett is set to the default input and output for sound
Tried uninstalling both apps, (using AppCleaner), cleared out the System Prefs and reinstalled. Didn’t help.
Went into Apps/Utilities - Midi and tried changing Sample rates etc but it didn’t make a difference.
Gave Logic a try again, worked flawlessly.
Rebooted the Mac, because why the hell not at this stage. Didn’t help.
Tried changing the USB-C cable on the Scarlett - Didn’t make a difference.
Tried changing the XLR cable on the Scarlett to the mic - didn’t make a difference.
Bumped up the gain on the mic and the interface to the max it could be. - didn’t make a difference.
Launched the Scarlett software (Focusrite control 2), gave that a try but everything seems to be set correctly.
The mic I am using is a Fifine AM8, so I disconnected it from the Scarlett and connected it to the Mac with the USB cable. Everything works!
So clearly the problem is the interface but why it only affects AI voice based applications I don’t quite understand at the moment, as every other type of audio application works. I’m assuming that the AI apps aren’t doing anything except checking for the default system audio and using that, so I am confused as to why it’s not being picked up.
Anyone out there have any similar problems and able to shed some light on this?
