Tsarnik

I would be petrified if Spotify finally licensed MQA. That way it could reach a critical of mass of young people who tend to replace their hardware rather often and hence could help expediting the dissemination of MQA-infested audio chips without the need of audiophiles, all agog with the sharpening of their blurred Diracs.

From the specs that I saw over the decades, DAC chips' onboard slow filters are usually linear-phase but exhibit a higher passband ripple than the chip's standard sharp filters. (Data of a 20-year-old chip and a popular modern one attached) I plan testing my DAC's listed ±0.003 dB passband ripple 🤬🤬🤬 according to the paper (–72 dB at ±0.7 ms and at ±1 ms) with some Red Book tracks. What I can say already is that using different filters on pop music tracks with energy up to fs/2 leads to several orders of magnitude greater variations in the waveforms near transients through filter ringing than the above mentioned echoes do. Messing with the soundstage, on the other hand, was a non-linear phase filter I tried some years ago: Archimago's intermediate phase suggestion https://archimago.blogspot.com/2018/01/musings-more-fun-with-digital-filters.html , as promising as it had looked to me and far from being a minimal phase filter, altered the soundstage too much for my ears.

You have got me intrigued, Shadorne.

I believe the authors tried to equate the passband equiripple with the effect of a https://en.m.wikipedia.org/wiki/Comb_filter, That seems to be rather far fetched to me, though.

Let's quote the conclusion as well, shall we: “However, this is fully under the control of the producer. Even a very steep linear phase filter with long ripple tails will not add to whatever ringing, if any, is present in the distributed file. If the goal is to avoid ringing from DAC filters, it can be achieved entirely in the studio. Nothing needs to change at the playback end.” Of course, everything is (and can be) more relaxed if one can use higher sampling rates and I certainly prefer a hi-res mastering if available. But 44.1 kHz is still out there as the most used distribution rate and it is on CD collections and that was and is my main concern. Anti-aliasing filtering needs to happen somewhere before the 44.1 PCM is written down. The microphone capsules are just the first in line. And a meaningful margin of low energy before fs/2 on my CD will require a rather steep filter, of course, but that will ring because of high energy around its corner frequency**, so we need not worry about the glockenspiel's HF energy above that (but about the microphone mentioned some posts earlier). Let the good producers decide where and how much some inevitable ringing shall be baked in, especially in classical music productions and leave me, the consumer, with a decent margin as quiet that my (hopefully steep enough) anti-imaging filter will have room enough to do its work silently (and maybe even afford to be a half-band filter). [But not: “Let the stylus clean your record.” (Linn LP12)] The bad stuff that can never be fully reconstructed either because of too much energy at my sharpest filter's corner frequency (maybe even 😇 at fs/2) or baked-in aliases (a no-no for me) will get the slow and/or the apodizing filter treatment. My 7½-years-old notebook remains just cool enough as to deliver me fan-noise-free (an absolute must) one-step ×16 resampling (Resampler-V) and DSD128 with the available SDM D modulator (no S&H) in Maksim's plugin. I have seen plots to know that your modulators perform better and your filters are better and more versatile. If money wasn't an issue I'd have a Qubuz subscription, a HQ Player, a new notebook, a new DAC and a new flat where I actually could listen to music without worrying about my neighbours. The only advantage I get from those thin walls is that I experience virtually no deep room modes - the waves just propagate through all my walls, floor & ceiling (except the outer wall). 😁 PS: A demonstration of a DAVE many years ago is the most memorable digital audio experience I've had until this day. But a certain Finnish software hasn't got that chance with me yet, so... 😉 Cheers ** in its transition band (thanks for posting, fas42)

Måns' explanation is indeed the most elegant one. I'm happy each time I this margin on a ripped CD track.

There's nothing wrong with a couple of Decibels Diffusfeldentzerrung, but the Montagnes Russes of the Sanken above 10k look quite omnidirectional. (^_-)

That is just not the case. Have a look at the microphone parks of the big studios: https://emil-berliner-studios.com/studiotour#equip-11 https://s3-eu-west-1.amazonaws https://www.thisismetropolis.com/studios/backline/ https://www.gabrielrecording.ch/equipment/ The typical recording mic reaches 20 kHz. Schoeps offers a frequency "extended" version of its amplifier, Sennheiser's MKH 800(0) series are also using EQ while keeping the small diaphragm condenser low-noise. Those mics are still the exception rather than the norm, even if the MKH 800 is over 20 years old. The CO-100K is 12 dB hotter at 22 kHz than at 1 kHz... For the sampling theorem to work the signal must be band-limited. The worst inter-sample overs/peaks I do experience with YT AAC (128 kbps). But a 11025 Hz sine at the right phase will be correctly (band-limitedly) sampled at +3dBFS (fs = 44100 Hz). 🙃 ISP's are a real nuisance but at least they can be easily dealt with by gain in a suitable audio player software.

Fair enough. MQA rulez. Not. 😁

Most of my classical CDs are good quality. Measurement mics aren't used for music productions. I don't know of any recording engineers force-feeding their ADCs 100 kHz signals. Often enough, "duller" microphones are being used for e. g. trumpets so that not much EQ'ing is needed out of the box (in jazz and pop productions). A carefully band-limited transient will have negligible energy for a nearly rectangular brickwall to start ringing. Dirac impulses and step functions OTOH are illegal signals. I have seen them too and they better receive some special care if they occur. 🙂 I'd divide the 90+% figure by four for my own CD collection.

A convolution of a Dirac impulse or a step function (think e. g. of a digital editing discontinuity) with a filter will show that filter's properties: a steeper filter usually has more taps than a more relaxed filter and so the response is longer (usually tapered by a window unless it's a fancy filter with a huge number of taps). But those two signals have no business in being part of a PCM music file to begin with.

One of the most important flaws of MQA is the resampling of 96 (88.2) kHz PCM to 48 (44.1) kHz PCM with a leaky filter in the encoder stage that causes aliases to fold back into the baseband and hence remaining in the file ("baked-in") forever. Insofar, the discussion about aliasing may inform some people about that issue. Especially, since aliases and images do get mixed up by some. I would also like to point out that a good Redbook mastering will contain no baked-in aliases, will have enjoyable transients but no step functions or Dirac impulses whatsoever. As such, it will have much reduced energy close to and at fs/2 so that the filter required by theory to perfectly reconstruct such bandlimited signal, the steep brickwall filter, will also practically reconstruct the original waveform perfectly without any ringing – or "blurring", for that matter. Hence there is also no need for any "de-blurring" which is a term coined by MQA, isn't it? So, I would kindly ask not to throw out the baby with the bath water. 🙂

I have made some more screenshots. A necessarily imperfect comparison between a Redbook source with lots of energy near fs/2 and two resampled files (8x), using two different filters. One can see how the steep filter Alt 9 has some energy at the very top that is preceding the transients. It can be seen that also filter Alt 2 is ringing at its (lower) corner frequency, but for a shorter duration. A better mastering with a nice steep attenuation before Nyquist would spare us to choose either sacrificing some high end for making the transients possibly more enjoyable or to enjoy the full band and maybe hear some strange sounding transients instead – with a better mastering we would just brickwall-filter everything Dave-style and be very happy. But I'm afraid there's no such thing as a carefree CD collection. Cheers

Here you go. In the middle window you can even see two back-folded overtones on top of each other that came from the forbidden zone beyond Nyquist. Nicholas

Use a spectrogram. And on music. It's actually quite straightforward. :-)