Audio Hijack 4.3.1
Rogue Amoeba has released Audio Hijack 4.3.1 with tweaks to its recently added Transcribe feature (see “Audio Hijack 4.3,” 6 November 2023). The update makes several backend adjustments to the beta Transcribe block to improve speech-to-text results, adds a language selector to focus transcription efforts on a specific language, adds support for up to five input connections, and displays a performance warning on all Intel-based Macs. ($64 new with a 20% discount for TidBITS members, free update, 37.3 MB, release notes, macOS 11+)
After a few years of hemming and hawing, I picked up a license to Audio Hijack. (I already have licenses to two other tools from RA)
The transcribe feature tipped me into the gotta have this column.
I had not updated my ACE installation (the low level interface with system audio) since before it required special permissions in the boot settings. I went ahead and did that and tested the transcription block out.
My first test was to run an entire podcast episode through the transcriber, then paste that into chatgpt and generate an essay based on it. Then have chatgpt provide what was left out or critiques of what was described and then integrate that into the prior essay.
It worked beautifully. This could all be done using other tools, but the transcription for meetings across a bunch of different conference apps is a very valuable idea!
There was a catch though. I found that my system’s audio selection became pretty unstable after I installed the ACE kernel extension. I could resolve it with a reboot but the symptoms were beachballs and confused state between Airpods and System Audio, eventually putting system audio into a broken state that required a restart.
I had to uninstall ACE to find the floor of stability again. I’ve got a cleaner system I also installed the ACE onto also running Sonoma, so I am hopeful to see if some other unusual config about my current machine was the source.
I also reached out to RA reporting my experience to see if there were any known issues being worked out.
I’d be curious if anyone else is using the latest MacOS and has the latest ACE plugin installed and actively toggling applications sources (Both pure audio to audio/video playback / webcam etc and toggling output selection (airpods pro / max / laptop speakers) and coming away with a perfect experience.
I got feedback from RA that some of the issues I’m seeing are unlikely to be related to ACE. Their customer support is solid. I’m going to toggle the extension back on and work with them to make sure its in a good space. Will plan to come back and provide some kind of update.
I am running the latest ACE and latest MacOS on a Studio M1 Max with 32 GB ram. I frequently switch between different outputs (wired speakers, wired headset, bluetooth headset) and with multiple input sources (wired USB mic, Teams, Zoom, Apple Music, web browsers). No problems and it is stable.
I wanted to follow-up that I have tested my use of ACE / Audio Hijack on a brand new M3 MacBook Pro and it has been completely stable. I believe that the issues I’m having on my M2 MBA are unique to that system and that I may need to strip it down to find out what is causing some of my media problems.
So, I to be clear, I do not think my issues on that machine are linked to ACE / Hijack. I will probably blow that install away and start fresh.