I’m sorry if people here treated you aggressively. Do you feel that I did so?
What are the physics in your view? Given that, within the band of frequencies that you are capable of hearing, there is no difference between 44.1kHz and higher sampling rates, what are the physical mechanisms that you believe allow you to distinguish between the two?
The reason for my scepticism is that you report hearing an
obvious difference between the two.
This raises a couple of red flags for me, for the following reasons:
- Even in those studies in which trained listeners (which you and your son are not) were able to discern a difference (and all such studies I’ve read have had methodological issues - although I’m open to the possibility there are other studies out there I’ve missed), listeners struggled to hear differences, could not hear differences most of the time, and certainly did not report hearing obvious differences.
- The majority of “high-res” files provided by streaming services does not in fact contain any high-frequency content (other than noise) that is not present in the standard resolution versions. So if you are capable of discerning differences, I would not expect you to be able to do so with the majority of recordings (on the basis of the higher sampling rate alone).
- Many high-res re-releases are also remasters, which obviously results in pronounced changes to the sound. That you have not investigated whether this applies to the content you hear differences in shows, in my opinion, that your claims are to be viewed circumspectly.
Can I suggest a way forward? Perhaps you could send me a high-res file of a high-quality recording that you’re very familiar with? I’ll downsample it to 44.1kHz and reduce and it to 16 bits with quality dithering, then place both versions back inside identical high-res “shells”.
This will ensure that:
- the original high-res file actually contains high-res content (I will check this)
- the standard-res file is properly dithered and filtered
- differences between masters are eliminated as a confounding variable
- you and your son will listen double-blind
You and your son can then attempt to discern a difference between the two versions using (I’d suggest) an ABX procedure. It should take no more than 15 minutes to run enough trials to achieve a statistically significant result.
What do you think?