>>105681677
The Denon perl pro does this and it's effectively just a randomised EQ. Behind the marketing BS it's using otoacoustics which is useless for this. It's great for diagnosing profound deafness in babies but useless for accuracy predicting ±6dB sensitivity at different frequency ranges.

What could actually work is measuring the reflection time of an impulse and using that to predict the ear canal length and resonant frequency but even then you won't auto fix it, you'll still need to adjust the EQ levels yourself