Slow performance for analyze.py and embeddings.py

[MZiegenhorn]

Hi there,

I'm running analyze.py and embeddings.py in concert (i.e., on the same dataset of 100k 5-minute files) and really struggling with performance.

Each file is taking ~150-200 seconds to run, which is a big issue with the number of files I have. I'm running the CLI version on a standard Macbook Pro (2017), with an overlap of 1.5 seconds (otherwise all settings are default).

I'm wondering if there's any way to speed up performance. I know a lot of the issue is that it's running on the CPU i/o GPU (thanks, Apple...). But have any other Mac users had this issue or found any way to speed things up? I'm wondering if it might be worth purchasing a Windows machine to use for this instead...

Thanks, BirdNET community! :)

[MZiegenhorn]

transferred my workflow to a Windows machine (desktop, way newer, etc) and it improved speed by an order of magnitude. I'm still interested in what sorts of speeds others are getting on Mac, if anyone wants to comment on their own setup.