How has everyone trained their ears to transcribe songs?

Currently I am separating songs with Demucs and then using anthem score to help identify notes when I am screwing it up.

I am just curious what others are doing.