The usual way people try to improve their ear is to mix a lot and hope something sticks. It works, but very slowly, because in a real mix every variable moves at once: while you're deciding the vocal EQ you're also judging its level, its reverb and its place in the stereo field. With that much in motion, isolating what's hard for you is nearly impossible.
Isolating one variable changes the pace of learning. Ten minutes identifying frequencies alone, with immediate feedback on whether you were right, teaches more about EQ than a whole afternoon moving bands without knowing how you're doing. It's the same principle behind a pianist practising scales apart from pieces.
And it's worth saying what this doesn't fix: hearing well isn't mixing well. You can identify a resonance instantly and still not know whether to cut it or leave it because it gives the track character. That second part is judgement, and it's trained by deciding, not by listening.
With ten minutes a day, most people start telling adjacent bands apart in two or three weeks. What doesn't work is the marathon session: perception saturates in about fifteen minutes.
They do, and to start they beat badly placed monitors in an untreated room. What matters is using the same ones for a stretch. What doesn't work is a laptop speaker.
The entry exercise for each skill is free and needs no sign-up. An account only carries progress across devices.
It removes half the problem. Hearing well isn't mixing well: you can identify a resonance instantly and still not know whether to cut it. That second part is judgement, trained by deciding rather than listening.
The costliest mistake is training what you're already good at. Before choosing where to start, take the mix skills test: eight questions, no sign-up, broken down by area. The method and the four-week routine are in the ear training guide. Before training anything it's worth knowing whether your listening system is lying to you, and that's in the monitoring guide.