Audio Classification | Analysis Performance Impact

First off, thank you for the new sound classification feature. It’s genuinely impressive and clearly a big step forward.

However, I did notice a major performance regression. Analysis time for < than 1,000 audio files on my system increased from 20 ms in the previous version to up to 2 minutes in the Beta. That is a 600,000% slowdown, which is substantial.

The test was done with fewer than 1,000 audio files, and even at that scale the slowdown is already very noticeable. Many users may have libraries of 10,000+ sounds, where this becomes a serious workflow blocker, especially when frequently reorganizing and updating sample libraries that require many folders to be rescanned.

For users (myself included) who would rather preserve fast analysis performance than enable categorization, it might be worth considering making audio classification optional. A simple toggle to enable/disable the classification step would allow us to choose what best fits the workflow, without taking away from the excellent work you have done here.

Thanks again for the great work and for being open to feedback 🙂