Please forgive me but suggesting FM as some reference for acceptable audio quality is not persuasive. I'll take my lectures on psychoacoustics from published experts, thanks.
I chose 10 kHz as an example, overall the crosstalk is very poor. I also said "dangerously close to audible", not a glaring issue.
Yes, at high volumes (not 1 watt, try 20 watts and 200 watt peaks) -60db is audible. Try it at home. Amir suggests -115 dB as the threshold of audibility for some measures.
From the AP website itself:
"values of crosstalk of 60–70 dB and more are desirable" - sounds like a minimum standard, not a goal.
I believe you are making my point, probably without realizing it. The masking effect is even more prevalent at higher SPLs than at lower ones. The ear-brain system will be preoccupied processing higher SPLs while masking out the simultaneous but much lower-by a factor-of-a-million(!) xtalk sounds.
Yes, I used the FM stereo example as an extreme one to let one know that you do not NEED -100dB (or better) crosstalk figures to get the full stereo effect (HD FM radio is not much better). Desirable, yes, ofcourse! But psychoacoustically unnecessary. Naturally, I'd like to have the best technical specs possible for all my electronic devices, if not to make me feel better. That's why, for example, people here frown upon devices with, for example, a 0.01% THD while praising 0.001% (or better) spec'd devices as being far more transparent....yet they ignore the big sources of distortion in their systems ... their main speakers which have single-digit distortion specs or worse at increasing SPLs. A good subwoofer, for example, is said to be transparent to the ear if its distortion is under 10% at high SPLs, with 5% being a really excellent figure. My MartinLogan electrostatics measure at 0.1% in the midrange and this is a superb figure...by speaker standards, that is.
AP's target of 60-70dB is an engineering target for the devices their measure, not one based on psychoacoustics. As an EE myself with 20+ years of low-speed but ultra-quiet designs and high-speed (10+ Gbps) circuit design experience as well, I'd prefer to set a goal of -100dB crosstalk or better across the full audio bandwidth, not because my ear can hear it if it's less, but because it lets me know I designed my product extremely well with careful PCB stack and layout, judicious use of targeted shielding, proper selection of capacitor values, etc. Pride and bragging rights instincts are strong amongst engineers. Besides, marketing would be screaming at me when they insist I have to exceed our competitors -100dB xtalk figures, with -110dB or better...a numbers game with these people. They'd categorically reject a -60dB xtalk design as being utterly 'terrible' and not competitive, despite it being audibly innocuous. Guess who wins this battle?
Have you looked at full bandwidth xtalk specs of typical 7-channel multi-channel power amps (e.g 200W/Ch typical), all sharing one power supply with their individual power cards crammed next to each other into one chassis like sardines? Maybe you shouldn't coz you won't like their xtalk specs at full power
.