Online it says “Its main advantage is that it removes most of the ringing that has been introduced upstream in the recording process when the original material was recorded and mastered.”
But if ringing is only in the ultrasonic range what’s the point of it?
The “point of it” is pretty much the biggest point in audiophile digital audio products these days, marketing! Much of the modern audiophile digital audio marketing employs the strategy of taking some potential problem with digital audio, typically either: Complete BS that never exists, something that was only a theoretical problem but never an issue in reality or something that was an actual problem during the development of digital audio recording in the 1970’s and early 1980’s but was solved shortly thereafter, and then misrepresenting it as a real, current problem. This allows them to market their audiophile product as a solution to this non-problem. Jitter is a classic example, it was an actual issue in the 1970’s but by the time digital audio was released to consumers it was already well below audibility in most consumer devices. By the mid 1990’s jitter was around 1,000 times below audibility in some/many devices and even the very cheapest consumer products typically produced jitter around 200 times below audibility. So a completely solved issue by 30 years or so ago but still commonly misrepresented as an actual problem today in audiophile marketing/reviews.
It’s a similar story with filters. Digital audio was really the first available mass market digital product and in the early 1980’s the processing power of available chips was extremely low (and the software/firmware very rudimentary). So much so, that in some processing areas it was more effective to do it in the analogue domain. Some early pro audio ADCs injected analogue noise to dither for example and Sony’s multi-track digital recorders didn’t have any digital filters, they employed a complex cascade of analogue filters. Of course in the mid/late 1980’s digital audio take up really took off, along with other consumer digital products (video games, office and simple home computers for example) and the development and production of processing chips exploded. The bandwidth and processing demands of oversampling and adequate filtering (which made digital filtering a “no brainer”), became available in the latter part of the 1980’s and during the 1990’s was pretty trivial!
I agree that the super sharp linear phase is the correct filter and the most accurate but 90% of my CD collection is not well recorded it’s compressed rock and metal, in that case I think apodizing or even minimum phase is a better choice for me.
Why would you agree a super sharp linear phase filter is the correct filter, why would you think compressed rock and metal is not well recorded and why would you think an apodizing or minimum phase filter would therefore be better? The only reason I can think is that you’re a victim of audiophile marketing (and/or those repeating it). Referring to the above, what is described as a “problem” or specifically “
The problem in practice” is actually NOT a problem. In fact, it’s arguably not even a problem in theory, let alone a problem in practice! Take the posted example of the Dave+MScaler, the attenuation of the freqs above the Nyquist Point (22.05kHz) is roughly -160dB, is that better than a more typical filter which only attenuates to say -115dB? Answer: No, it’s actually exactly the same! In practice of course we are not just measuring the analogue output of the DAC with an AP555, we’re listening to sound, so at the very least we’re constrained by the theoretical limits of sound itself. The absolute theoretical limit of sound is about -23dBSPL, IE. If we had a hypothetically perfect anechoic chamber where no sound could enter from the outside and no sound was produced inside, we would measure -23dBSPL, which is the noise produced by the random collisions of the air molecules inside the chamber. There cannot be sound lower in level than that (unless we remove the air). So let’s say you are listening to your music at a moderately loud but not potentially ear damaging peak level of 85dBSPL, our ultrasonic content at -115dB would therefore be at -30dBSPL, so it cannot exist as sound, in fact it’s roughly half the level required to exist as sound (6-7dB lower than the -23dB level of no sound). The -160dB of the MScaler would be at -75dBSPL, about 52dB lower than required to exist. So, they’re exactly the same, in both cases no ultrasonic sound exists. This is in theory, it assumes hypothetically perfect speakers, amp and anechoic chamber, in practice the situation would be even more silly!
Regarding your other points, most rock is at least decently well recorded, unless it’s very amateur recordings you’re listening to but even then there shouldn’t be much ringing as even cheap prosumer ADCs generally have good filters these days. If it’s heavily compressed during mixing or mastering that will only rarely result in ringing and when it does, it’s of a low magnitude and virtually all in the ultrasonic range anyway (unless it’s deliberate). There’s plenty of other things to be worried about that are audible! The best choice will just be the default filter (the fast linear phase one), there almost certainly won’t be any audible difference between other filter types unless they’re particularly silly, for example, with a roll-off well within the audible spectrum (I‘ve seen some start at around 10kHz), in which case they’ll be worse/lower fidelity.
G