Just saw these Bob Stuart comments from Stereophile, responding to an article by John Atkinson called “Zen and the Art of A/D Conversion.” Not sure whether it helps or hinders the conversation here!
Hi Dear Roon Fellows,
I’m happy to have vacation for a little bit more time and I think that maybe I will extend it
I following the “MQA Disappointing” discussion as I receive notification on my email, I must say that some comments are not especially kind or necessary !!!
But it’s different from my view of course, it’s different for you who are in the discussion !!!
Be kind to each other my Dear Roon Fellows
Love & Respect
Your son is a lucky kid getting the gift of good audio from his dad. I agree about not discounting MP3 users–we all had to start somewhere (transistor radios…), and after all, the important thing is people listening to music, not how they do it. As for vinyl vs. streaming, I guess my implicit deal as a streamer is trading off a little sound quality for the fantastic world of new music at my fingertips–sheer magic in my opinion. TO MY EARS the SQ difference between the two is considerably less than often stated, based on listening to the two methods, and it is decreasing as technology improves (given false starts along the way). Your point is also a good one that a claimed improvement in SQ from any method may cause improvement in recording from the outset. Maybe the decisive factor in all this is corporate profit–when (not if) some kind of method of improving SQ becomes universal, the profit potential will be unlimited, due, again, to the sleeping giant of the public’s desire for better sound, and companies will race to implement it. It’s gonna happen, with bleed into audiophile audio. The future is unavoidably bright.
Yes, I’ve read Stereophile latest issue on this topic. You noticed John Atkinson covered on different types of filters and he seemed to favour the slow minimum phase type but he failed to mention about the aliasing effects (weak filter) and non linear distortion at high frequencies that may affect the music we hear. It is one sided review obviously to promote MQA.
He quoted: ‘Charley Hansen was a vociferous critics of MQA. However, if you compare what I’ve written in the paragraph above with my earlier description of the goal of MQA’s developers, the are essence, identical!’ If Charley Hansen is still alive, I don’t know what’s his reaction to this…
Well, he is missing one important point here. What Ayre does in recording and listening mode does not involve any form of lossy compression in the link. I think he needs to explore 2 variables in MQA;
Does the lossy compression artifacts affect how we hear the music?
Does the de-blurring filters used to correct tempo blurring which also create artifacts like aliasing and non-linear distortion at high frequencies does this affect how we listen the music?
Until all these effects can be put into precise technical graphs (still waiting 4 years after its announcement) and debate along with PCM and DSD, people is less convinced with just claims after claims. I’m utterly disappointed.
Just a question: Have you listened to any MQA files through a compatible DAC? If you have and still don’t like MQA, than that’s fine and I respect your very wordy opinions. But if you haven’t even listened, then…
Yes I did that. I have (regrettably) a Node 2 which is fully MQA compatible. Now, ignoring his general sound quality (which in my opinion is a joke) there was nothing in the MQA files to make me say wow. Some of them sound better that the CD version (both TIDAL), some don’t. For most of my music, I still get the best sound (in this configuration) from my flac files (now, I can’t say that the TIDAL MQA, the TIDAL flac and my flac are all having the same master at the origin). Comparing with 24/96 or 24/192, MQA brings absolutely no audible improvement, on the contrary. Going even deeper and testing vs. DSD the MQA just disappears…
Please note that I don’t pretend that my test are professionally made or extremely acurate. They reflect my experience, but that’s the only thing that maters to me. In the end I consider MQA being another expensive joke on us and is a no go for me.
Now there is another aspect of the story. Being so disappointed with the Node 2’s sound, I am now using it just as a network bridge between my server and the coaxial input of my SACD Player/DAC which sounds (again to my ears) way much better. Obviously I’ve lost the second unfold in the proces and the MQA still doesn’t bring music to my ears.
My setup (the Node 2 especially) is temporary, I just wanted to have something to test for my ears (at that moment ROON, TIDAL and MQA were all new for me). I’m doing my homework now for a new DAC, but one thing I’ve learn in the process is that I really, really do not care if it is or not MQA compatible (which narrows my search to the ones without MQA since I don’t want to pay for something that I don’t care about). My library has 2 or 3 MQA albums I bought for tests and a few more from TIDAL because at this moment they do not have other versions available. And it will stay this way.
MQA sounds super sweet here on My Bluesound Pulse 2. Certainly less fatiguing over extended listening over CD files. I always look for an MQA Version now.
CD files sound great also and I can see myself listening and enjoying them forever as a lot of my music will possibly never be in MQA, certainly my bootlegs lol
But this lack of fatigue draws me in…
Looking for some listening between two real speakers will probably bring you more musical enjoyment than a format or another…
I have real speakers as well. In the summer months we live in our conservatory and my Meridian set up is saved for the winter. I know what great music sounds like spending a lot of time filming live music in intimate environments. My most recent was The Lauren Housley Trio In said conservatory.
Live music is the best Hi Fi available IMHO.
We might not share similar views of MQA but I do like the way you mention different artists in some of your posts.
Have looked into quite a few and now own some of their music. Thanks for that!
From your description here, you site dissatisfaction with the sound quality of your component irregardless of the file type, so expecting MQA to solve the ills of your system is hardly a fair expection.
Further if you are not able to benefit from the second unfolding of the MQA process, you are crippling the technology and I really can not see how one can pass judgement on the process, really at all.
In my experience, the sound quality potential of MQA files are dramatically limited when just the first unfold is done.
You are of course entitled to your opinions, but from your writings above, I find very limited value unless your are listening to a system that you are happy with, at least as a base line, and also unless one is able to audition the formats full potentials.
Otherwise it’s akin to condemning HD television technology while only using a standard definition tv…fatally flawed.
You got it wrong. I do not expect anything form MQA. I didn’t even expected the format. Knowing the limitations of a system is not an impediment in assessing how different things (formats in this case) are sounding through that system, it is just a matter of experience.
My system is able to benefit from the full MQA, I just chose not to because there is no benefit (for me, my system and my musical enjoyment and sanity).
My post was a direct response to somebody here who asked me a direct question and my answer was not intended to have any other value. It is just my experience and has value for me only.
I sincerely (and respectfully) believe that you didn’t understood my post or you did not read it. Either way I can’t find anywhere your opinion/experience/whatever on the topic, so…
The ‘super sweet sound’ is probably caused by aliasing image and non linear distortion of the MQA filter (weak filter). In my experiment with NOS DAC ( No filter) playing back (also mentioned in the latest Stereophile by JA) with 16/44.1k, it does indeed sound a lot sweeter than with over-sampling filters. You may want to try out NOS DAC, it makes all your collection ‘super sweet sound’ without re-buying or invest in MQA version. Save $$$!
Thank you, the Music in whatever format is the main reason we are all here and so much great stuff will never be in MQA for a long time to come, if ever. MQA or not great music can and does sound great.
A case in point is the recording I have from my Lauren Housley birthday party was mixed, Mastered and made into a double CD for me in four hours as a gift for me before we went away for a break.
This became the sound track of the trip and a few tracks there gave and still give me goose bumps in the car with all the road noise and far from ideal conditions.
So yes, we can argue about formats, their value, their technologies, the motives of the promotors etc but we mustn’t loose sight of the Music…
If that’s the case, bring it on. It sounds sweet and I like it. I haven’t had to buy anything so far and not looking for upgrades with this sound in this space…
My theory is that the noise is high frequency and inaudible but it helps linearize poorly performing DACs. This effect was proposed as early as 2002. The idea is simple: high frequency noise helps randomize non-linearities in a DAC. This would explain why upsampling to higher sample rates often makes audible improvements. The HF noise isn’t audible but it ensures that the DAC is decoding the audible band in a different way due to superimposed HF noise on the signal - think of it as a form of dithering. The linearity issue is the extremely small voltage error between adjacent bit levels on the DAC and not an overall linearity issue - like quantization noise there will always be small errors but as we know with jitter - extremely small errors correlated to the signal can produce audible distortion.
Mathematically this makes sense. As you go to high sample rates with less bits then the non-linearities of the bit transition disappear. However, a high sample rate system depends on a super accurate stable clock - if any of the digital logic circuitry causes correlated shifts in the clock you have the same issue as that of a ladder DAC - tiny non-linearities between bit level transitions. The more you study these issues the more it is evident that problems can still bite you. The huge advantage of high sample rate systems is that they are cheap to build but they may have equivalent issues to old ladder DACs. LIM (logic induced modulation) is a huge issue. As bad as jitter, IMHO. Again these issues were researched by Ed Meitner decades ago but folks rarely read up on what he discovered and manufacturers for the most part ignore these issue and hope they go away.
So if HF noise like in MQA or hi-Rez files (anything above 30KHZ is inaudible) improves the sound of your DAC this may suggest the DAC has linearity issues.
I think you are referring to dithering; it adds some random noise to the upper bits to improve linearity. I was talking about aliasing image get reflected back to the audio range due to the use of ‘weak’ filter.
Aliasing has two effects, namely, in the upper band >20kHz, there’s substantial noise increased with no apparent subsonic music contents. If the recording do have some subsonic music contents >20kHz, it will be likely masked off by the noise! Second, this is more of a concern is the image get reflect back to the audio range causing intermodulation distortion with the music contents, creating an entirely new harmonics previously not available in the original music contents. In simple plain word, it will alter the music textures and harmonics.
That’s why in digital recording we stress the important of keeping aliasing image as far away from the audio band; over-sampling will shift to higher frequencies while linear fast roll-off filter will attenuate the image energy. There are two type of filters, linear and no-linear. Linear filter (fast roll-off and slow roll-off) has been used since the introduction of CD in the 80’s. These filters is linear up to the audio range and does not exhibit non-linear distortion. However, it does have a downside called ‘tempo blurring’; they have high degree of pre and post ringing.
The second class of filter is non-linear which also known as minimum phase filter; both fast and slow roll-off (MQA uses slow roll-off) are available. Because these filters are non-linear, they create distortion as the frequency comes up. Second, these filters are considered ‘weak’ when it come to attenuate the aliasing image. However, they do have improved transients response with no pre and 1 cycle of post ringing (slow-roll-off).
Dithering is done on LSB. I am simply making an analogy to dithering (the randomization of quantization noise) and saying that HF noise from ghost images etc could help randomize noise from DAC non-linearities (read the PDF article for more details).
I agree on linear phase filters and how upsampling (in D to A) shifts the ghost higher so they are easy to filter.
The amount of ringing from linear phase is for D to A is not of concern. Most ringing comes from mix and mastering where high Q filters are used on things like snare.
The aliasing issues of in band noise being folded in happens in A to D. A sharp anti-aliasing analog filter is required above Nyquist prior to digitization.
I think we agree on everything. Sorry if I didn’t explain clearly.
Here is an explanation of differential non-linearity (a big problem in ladder DACs but likely present in Delta-Sigma too because of logic induced modulation of the clock.)
The plot explains it best
As you can see both DACs in the graphs above are overall linear over the operating range show, however, only the DAC in plot A (on the left) is differentially linear.
Now imagine if most DACs behave like the right hand plot and have some systematic differential linearity issues. If we add some ultra high frequency noise (so high as to be inaudible) then the systematic errors that occur at specific bit steps required by the audible music signal will become randomized! This means higher sample rate files and hi-Rez files may sound better than red book simply because the ultra high frequency content whilst inaudible actually “dithers” the systematic differential non-linearity and makes the issue random.
Listening tests can be beside the point if a person objects on principle to some of the assumptions underlying MQA, personalities promoting MQA, etc.
For a parallel, I might find that I like the sound quality of a dynamically compressed loudness war remaster – but maybe I still shun the remaster because it is an adulteration of the original mastering.