A Cure For PCM “Digititis”
What do you think when you read statements like, “I’ve felt from the start in the 80’s that there was something inherently wrong with PCM audio? For some reason (and I’m not alone) I cannot stand listening to PCM recordings for more than 5-15 minutes without experiencing some sort of inner distress”. Do you believe this? I’m not sure why this nonsense persists. There it was in a review of a program called “Master Class”. Here’s how the software is described on the Daniel Hertz SA (Mark Levinson’s company site):
“Daniel Hertz Master Class is a music player software for Mac with revolutionary capabilities. For the first time in audio history, listeners can have the sound and feeling of SACD and the best analog master tapes from PCM digital sources including CD and compressed music downloads.”
I’m already on high alert after reading this. As the producer of PCM recordings with more dynamic range and higher fidelity than SACD (DSD 64) and analog tape, why would I possibly want to diminish the quality of my recordings by running them through Master Class? It’s crazy to think that this process or any other process that changes the sound of my masters would be an improvement. Since when did SACD and analog tape become the ultimate fidelity standards?
And here’s an overview of its features:
• The infamous A+ algorithm that smoothens the PCM step function waveform.
[NOTE: PCM is not a step function waveform…the output of your DAC is a smooth, accurate, reflection of the analog waveform that entered the ADC at the source end of the process! Maybe the algorithm is “infamous” because it’s snake oil.)
• A six band graphical EQ that let’s you refine the tonal balance (frequency response) of the recording and match it to your playback system.
And finally, the user manual explains the “concept” behind the “infamous” A+ algorithm this way:
“The analog waveform is smooth. The PCM digital waveform is a step function that only approximates a sine wave. The human brain, like the sine wave, is from nature. The PCM digital audio step function waveform is something made by technology that does not exist in nature. A+ is proprietary algorithm that fills in the ‘steps’ to smoothen the waveform so the brain responds more like if it were a sine wave. A+ uses the original signal to generate the ‘fill’ so there is no non-musical information being added. A+ adds no noise or filtering.”
Mark Levinson is a well know fan of DSD and obviously doesn’t understand how PCM digital encoding works if he believes that a “PCM digital waveform is a step function that only approximates a sine wave”. First, there is no such thing as a PCM digital waveform. The output of an analog to digital converter is s stream of ones and zeros that are stored on some sort of media. It is not a waveform…the stream is a collection of data. If the source is a sine wave, then the output from the DAC at the other end of the process is an accurate recreation of that sine wave. There’s no approximation…it’s like using a set of blueprints to construct a building. Is the resultant structure and approximation of the original design?
The type of thinking described by Mark Levinson’s website incorrectly employs analog signal thinking to digital methods…they are two discretely different ways of dealing with audio. It sounds like his “unique” process is upsampling or using interpolation to “fill” in the data stream with more samples. This technique provides the designers of the converters to use better filters but it doesn’t fundamentally change the fidelity of the recording.
I’m headed down to the Newport Beach Audio Show shortly…hopefully, I’ll see some of you there. And at 2 pm, I’ll be one of the guests on Scott Wilkinson’s Home Theater Geek show once again.
Audiophiles who continually claim that digital is fatiguing need to either fix their rooms or need to get different equipment. I have experienced far too many rooms with serious sound issues. I have also experienced systems that are put together to basically boost the upper octaves that the listener can no longer hear. Putting a good digital system in those rooms is unlistenable over long periods. But, so is the vinyl they play. This is really a crazy hobby,
There seems to be a similarity between some audio accessory promotion & politics. An apparent lack of knowledge & the belief that if BS is repeated often enough, it will become ” Truth”.
I’ve avoided the politics analogy…but I agree completely!
First I’ve heard of Daniel Hertz Master Class software. Is this another form of MQA?
No, this is a process to dumb down fidelity to match a certain expectation.
I really disliked 1980’s era CD’s but I think current CD recordings are significantly better and Mark’s recordings obviously the best possible with current standards.
I connected a $30 Dayton Audio DTA-1 Class T Digital amp to my 1980’s ADS L400 speakers using a $35 Chromecast audio adapter (AKM AK4430 192kHz 24-bit stereo DAC) the other day and I think it sounds awesome for the cheap cost and compared to my NAD integrated AMP and Sony CD player from 26 years ago. It must be pretty much CD quality at up to 320 kbps streaming. Fun!
I don’t know what Mark thinks (though curious) but I think it is safe to say that the at the dawn of the CD back in the late 70s/early 80s the technology still had a few growing pains to get through.
But by the early 90s, I think the CD sounded absolutely grand (and still does!).
NOTE: I try my darnest not to buy CDs. It’s is a dying format.
CDs were less than perfect at the dawn of the format. However, I do remember be amazed when I heard a Schubert String Quartet in the format…silence in between movements and notes. The format is not perfect, but close enough for virtually all commercial music.
Good grief!
It is amazing to me the number of geniuses who think they can fundamentally alter for the better the sound POST PROCESSED SOURCE by adding upsampling. I have never heard anything aesthetically pleasing in anything upsampled.
i think i’m starting to understand now what Salieri meant in his comment on Mozart’s music in the movie Amadeus – “TOO MANY NOTES”.
Trust Mark Levinson will at least try and clean up some of the notes.
Mark W thanks for this snippet of truth serum again.
I’m glad you used the blueprint of a building analogy. If the uneducated, misinformed tripe that passes for insight into the nature of PCM were true for the reasons they say, then any building built from a blueprint would be inherently blue and embedded with bad blueness, and no-one could walk into it for more than 10 to 15 minutes without starting to feel blue and depressed.
Very good
Same ole anti digital claims, and mis-representations of the technology, we’ve been reading for near 30 years now.
Don’t these folks ever tire of repeating these lies?
And then they come out with the latest “magic dust” to cure digital of it’s ills and lift a thousand veils, LOL.
If I were to just combine a small percentage of the magic dust cures that were claimed to revolutionized digital sound in the last three decades, when I fired up my stereo the walls would disappear and the performers would be transported directly into my listening room.
Beam them down Scotty. 😉
Hi Mark,
an interesting piece and perhaps you could help me here, is this (Mark Levinson) site selling software that converts to DSD then back to PCM to allow the A+ algorithm to work then back to DSD then on to the DAC for conversion?
I didn’t seem to understand fully how this software works or indeed it’s intention.
Wierd that respected engineers find convoluted ways to get it wrong, when if they stuck to the (relatively simple) hi res PCM path they would get the same if not better result.
Leaves me baffled sometimes.
Whilst I am on, any further progress on the MQA front? And any update on the book?
Particularly interested in your comments on MQA, as another site has used a Mytek DAC for with and without comparisons of Hi Res tracks and the verdict was……… it sounds better. Just to be clear I believe all the tracks tested were analog masters and the piece was clear, the improvement comes from “deblurring in the time domain”.
I will ask more when the time comes. Please continue to keep us informed of these sites/companies who are trying to pull the wool.
Kind regards
Gordon
No, they are simply processing PCM files to sound like analog or DSD. I’m busy with the book everyday. Been working on the Blu-ray demos lately…progress is slow but moving. I’ll get into the MQA thing very soon.
Oh my Mark. I see you’ve wandered into one of the most pretentious and laughable corners of the audiophile blogosphere with that review :-).
‘Master Class’? ‘A+’? Playing with the sliders allowed ‘90%+ matching of the vinyl’ sound (and that’s supposedly good)? A ‘strong trend and movement towards DSD’?
As usual, the dude’s clueless. Basically sounds like it’s just some kind of upsampling software with whatever DSP effect thrown it to make it sound “analogue”.
I’ve been there before but hadn’t seen the nonsense on the Mark Levinson software.
Archimago, it has got to be this.
Mark,
I wonder about how long one can spread those twists of facts any further.
It is so brazenly.
After more than a quarter of a century when CD was launched in 1982, there are still people who try to fool the world.
I can live very well without this kind of “Master Class”.
On the other hand it might be that the author of this nonsense has no single glimpse of what he’s talking about.
Then he’s just a fool.
Regards
No, I think the output of your DAC is NOT a smooth, accurate, reflection of the analog waveform that entered the ADC, but the “toothed” reflection of the samples density/frequency. Denser the samples – smaller the teeth. Smaller the “teeth” – smoother the recreated waveform. – higher the definition.
Yes, I think the point of high definition is denser samples (amplitudes) queue, and not the sound frequencies above 20 kHz. We must not confuse two different terms – sound frequency and sampling frequency. The sound frequencies above 20 kHz are wellcome, but ALL the sound frequencies can be higher defined or lower defined, depending on sampling frequency.
The output of my DAC is a smooth, accurate, reflection of the analog input…there are no “teeth” or “steps” reflected in the waveforms sent to the amplifiers. This notion is a falsehood that is perpetuated by those whose advocate for analog.
The reconstructed output of any properly operating DAC is a smooth, relatively accurate replica of the analog input signal that was processed by a properly operating ADC, not just yours. Theoretical accuracy is possibly compromised only by construction and materials. Mladen’s comment indicates that he(she) thinks the output is smoothed digital.
Your DAC doesn’t have any analog input. It’s input are the discreet amplitude samples (a queue of vertical lines on the horizontal time-axis). Their distances in time (the distances between vertical lines) cause “teeth” or “steps” on the curve formed by the vertical lines’ summit points. Smaller the distance (more samples per second) – smaller the “teeth” or “steps” – bigger the similarity to the original recording – bigger the fidelity – higher the definition. That’s the point of HD.
If your DAC gives a smooth curve, it only means that it fills the gaps between samples with approximations. So, if you make more (real) samples in the beginning, then there is less need to add (approximative) samples in the end.
Obviously, dear Mark, we are advocating THE SAME. I rejected my LPs and bought my first CD player in the year 1986. Now I have much much more SACDs, DVD-Audios, Blu-rays and mch hd flacs then CDs. But, it is also true that 6 ANALOG cables connect my Oppo 105 player directly to my Rotel power amp and my Mirage SW.
Mladen, I don’t really want to belabor the point but your understanding of PCM and digital signals is incorrect. There are data, yes. They are point in time that are bound by the sample rate and word length. The distances in time (the samples) don’t form teeth. They are simply data that once reconstituted into analog voltages cease to be data and become identical to the source input. The notion that more data chunks somehow smooths out the digital “waveform” is fundamentally flawed. The data words are used to reconstruct the analog waveform and within the constraints of the sample rate and word length, a good system outputs a completely accurate analog signal.
Mark, if sampling process doesn’t leave any empty time-intervals (picturesquely, curve gets teeth or steps), why do you need higher than CD sampling rate (96 kHz) at all? If it is for the sake of sound frequencies above 20 kHz only, then it means you reduce the whole concept of high definition (HD) to the super high sound frequencies only. And that would mean that “ordinary” sound frequencies are good enough in whatever sampling rate. I simply can not accept that.
You have the basic idea correct, I believe. The move to 96 kHz/24-bits increases the frequency range, allows smoother filters, lowers the noise floor, and moves pre-ringing and “temporal blurring” to near 48 kHz, where you and your equipment won’t hear it. My point is that there are no “teeth” on the output….it is a smooth analog waveform.
Mladen – this video presents a more in depth discussion of the stair step myth.
http://productionadvice.co.uk/no-stair-steps-in-digital-audio/
I think with USB the waveform is like jagged broken teeth is it not? You need to get linear USB power supplies and special cables to smooth it out don’t you?
Please note dripping contemptuous sarcasm.
Then go watch this video which will forever put the lie of stair stepped digital to bed. Using all analog source and monitoring gear.
https://xiph.org/video/vid2.shtml
Mladin,
The magic is shown here with the key point at around 7:09 through 7:58.
https://www.youtube.com/watch?v=FLPqshreE-g
All properly executed DACs include a reconstruction filter as designated Hr(omega) or hr(t).
Hello Mark
With all the advances in digital audio over the past 10 to 15 years, there really is not much more improvement to be had. What’s the poor snake oil salesman to do?.
One must wonder with this magic software.. does it need to burn in?. Also are the bytes they use to make that software oxygen free?
Hi Mark, it was truly wonderful to meet and talk w/you on Thursday.
Barry has been working closely w/ M.L. for years and Mr.L came up and spent a long day into night w/me too about 2 years ago ,I heard the whole sermon, then spent 2 hours listening to I-tunes AIFF files that had been run through master class. In the end, a simple Johnny Cash all-acoustic CD on a good mid-priced player sounded like first gen compared to the massaged files. But like so many ‘lifers’ in Hi-Fi, I highly respect him just as I respect you whether or not we agree on sonics or philosophy or whatever.
It was a sad truth to share w/ you how the public is now apparently averse to truly dynamic recordings, and what enormous misunderstandings exist simply about the meaning of the word ‘dynamic.’ The one thing that is so often ignored in audio , ” Less (loudness in this case ) is sometimes more”. I will be looking to see how your AC cord investigation goes. Only one question: Is it possible that we can hear things we haven’t yet learned to measure?” Best, Craig
Craig, I enjoyed chatting with you and Barry. It was great to finally meet. Cheers
Craig, Of course it’s “possible” that things can be heard that we haven’t been able to measure yet.
As soon as the golden eared “trained” listeners start submitting to bias controled DBT’s, we all can begin to zero in on the things that are truly being heard. It’s so hard to separate the enormous number of flights of fancy that have taken over audio land today from what is in reality a very tiny number of real listening reports.
If those $6,000 2 meter power cords can truly lift a thousand veils and the differences can honestly be heard in the hustle and bustle of a audio show, then you’ll be able to identify the differences when the lights off and the listening is done under controlled DBT test conditions.
Cheers,
Sal
Just a thought… If someone has this software, should try running a test signal through and see what it does to the frequency response and dynamic range. Could also run an impulse response through and see what kind of digital filtering is being done!
Hey, if anyone has the software, contact me at the Computer Audiophile with a PM and maybe we can give this an objective run-thru!
It would be interesting to see what sort of digital processing the software is doing to make things sound “like SACD or analog tape”.
The only people foolish enough to challenge the fundamental knowledge given to us by scientists, engineers, and mathematicians like Claude Shannon are audiophiles and the people who manufacture expensive equipment for them. From my point of view and I’d bet from that of ever serious scientist and engineer in the field they are a joke.
https://www.youtube.com/channel/UCJIjs5B0HQtbuH48uibfLqQ
Bell Telephone Laboratories is where the modern world of communications was born and continues to develop. It made more important discoveries and inventions in the field than anyone else and is still going strong. I know, I work there.
Yes, Claude Shannon was a genius, and yes, Shannon’s Law along with the Nyquist theorem are critical components in a whole lot of today’s communication technologies, but they are not the last word. You may want to read up on the Cheung-Marks Theorem whereby “reconstruction error with unbounded variance [results] when a bounded variance noise is added to the samples” and quantization error “quantization is a many-to-few mapping, it is an inherently non-linear and irreversible process (i.e., because the same output value is shared by multiple input values, it is impossible, in general, to recover the exact input value when given only the output value).
Quantizing a sequence of numbers produces a sequence of quantization errors which is sometimes modeled as an additive random signal called quantization noise because of its stochastic behavior. Of course, the more levels a quantizer uses, the lower is its quantization noise power. You may want to refer back to Shannon on channel capacity for some clarity on this point. It should also be remembered that these errors are cumulative and express themselves as greater noise. Shannon again: C= B log2 (1+ s/n)
But there is more. Jitter – the variance in arrival time can be an insidious issue in any digital system. As can ‘ragged edge triggers’ most noted in cheap RAM chips that vary not only in their rise times, but also in the latency as to when the rise time begins and ends. This leads to bit-level errors that raise hell with quantization. And again, these errors are cumulative, so that by the time the signal has been processed, re-processed, de-processed and re-processed again … and again, it becomes a dancing bear. The miracle being not that the bear dances well, but that he dances at all.
The real problem between the objectivists her and the subjectivists is this: objectivists presuppose conditions of the system absent these grimy details because they litter the beauty of the underlying math, and therefore do not exist, or if they do, do not matter. Subjectivists lack the objectivist math and tools to characterize their experience – and are subject to a lot of frankly nonsensical ‘lifting of veils’. But just as being paranoid doesn’t mean they’re not after you, being a subjectivist doesn’t mean you’re not really hearing things.
And in my final nod to Dr. Shannon, a corollary, valid in digital electronics, valid even more in our current domestic environment: The more noise introduced into a system, the more error correction the system requires. When the noise level = the signal level, channel capacity reaches zero, and communication stops.
This is a scope image of an iFi IDAC2 playing a 1 kHz sine wave with the “filter” switch in the “bit perfect” position. My advice: don’t use that setting.
https://goo.gl/photos/STjiSrw8T3S4q1ky6
Really? This is amazing! The analog output of the iFi IDAC2 should be a smooth analog waveform.
The “bit perfect” setting bypasses the DAC chip’s internal upsampling with simple sample repetition up to 352.8/384 kHz, which is then processed by the sigma-delta modulator. I can’t fathom why they included this option, but I’m sure there’s someone out there who swears it sounds better, oblivious to the horrendous distortion.
Alan, sampling process leaves smaller or bigger empty time-intervals (picturesquely, curve gets teeth or steps).
You can’t get smooth curve reconstruction unless you add approximations of missing (skipped) data samples. It is far better to make many real samples (HD) in the beginning, then to add approximate ones in the end. Our respected dr. Mark implements it in practice, but denies it in theory.
Mladen, take a look at the videos that have been offered. A PCM digital system outputs a “smooth curve” without adding anything. Of course, there are limitations based on the Nyquist-Shannon Theorem but if you live within the science AND engineer things properly, you can get great results.
OK, it seems smooth curve restoration doesn’t need high density sampling but good calculating and filtering algorithms. High density sampling obviously serves only to reach sound frequencies above 20 kHz. At this point I am ready to give up theory discussion and return to my practice – listening to HD music. But, back here a quite disturbing practical question waits for me – Why do I like HD music more than CD (to say nothing of LP), when I am 63 years old man and certainly don’t hear 20 kHz (to say nothing of ‘above 20’)? Is HD then really a mere autosuggestion?
The higher sampling rate can have an effect on the “audio band” and it makes the design and implementation of filter easier. It also moves pre-ringing and temporal blur out beyond audibility. If you were to compare a CD version and a High-Resolution version of a well-recorded tune made, it would be very difficult to tell them apart.
So, if HD and CD are very difficult to tell apart, why do we need HD at all?
Because recording and producing new recording in HD provides the potential to meet or exceed the capabilities of human hearing. Engineers now have the possibility of increasing fidelity even if few productions take advantage of the new capabilities of HD audio. Most of the commercial stuff you can purchase is just hype and sales gimmicks. But with a great system and a great recording, it is possible to exceed CD fidelity…for most people it’s no big deal.
All I want is for my Benchmark DAC2 to receive the zeroes and ones to do its magic. The function of any digital audio delivery software is to deliver it losslessly, something that JRiver and others already do in fine style. The rest is just snake oil when it comes to digital.
I’m sure that most critics of digital have never heard it in extremis, as it were. Put another way, since my Benchmark 2 came into my life a few months back, the turntable hasn’t been used at all. I’m hearing music that’s been hidden in CDs I bought over 30 years ago. CD is an astonishingly capable format with the right DAC investment.
Thanks Curtis. The critics of digital are stuck in the analog world and want to use analog descriptions to justify their preferences.
One of the things I really love and enjoy about Marks site is the number of highly trained and experienced people who post here, including the folks who take issue with some of the subjects. The thing that really caught my attention was the 6 band EQ! When I see an EQ on a consumer playback system I’m immediately thinking something is wrong with component(s) and/or your listening area is a disaster. Whenever I see a household system with an EQ it’s rare that any levels are attenuated, nothing at or below flat. Attenuation never seems to be an answer. It just occurred to me that the EQ knobs are usually in a stair stepped kind of arrangement-maybe I’ve unwittingly discovered a point of confusion 🙂
I stick with 96/24 downloads is enough for me and apparently if you go for higher resolution the noise floor is increasing .I was into LP way back in the 70/80 now I stick with CD’S and downloads .In regards of MQA some people that had some listing testing were impressed ! However it will take a lot of time to produce enough albums that it can survive otherwise it will fail ! For the propose of streaming it has potential as the file is not large !
Sony and AVS Forum have teamed up to define and answer questions about HRA.
Get Ready for Ask Me Anything: Defining High-Res Audio
High-resolution audio (HRA) is a hot topic these days. It promises better audio quality than MP3s and even uncompressed CDs and equivalent digital files, but there’s a lot of confusion and misunderstanding about it within the ranks of audio consumers.
To help address this problem, AVS Forum and Sony are offering AVS members an opportunity to ask an expert anything they want to know about high-res audio in a series of real-time interactive sessions. The first session will be focused on defining high-res audio—exactly what it is, how it’s created, where to get it, and what you need to fully enjoy it.
I think you could have some fun with this.
http://www.avsforum.com/forum/91-audio-theory-setup-chat/2550473-get-ready-ask-me-anything-defining-high-res-audio.html
This does sound interesting. I called my good friend Scott Wilkinson at AVS Forum and asked him about it. He texted me back that he’s on vacation but would reach out to me when he returns. I’m going to write my blog post today about the continuing marketing effort by the DEG, Sony, labels, and others to push a failing concept. I think most people have figured out that high-resolution audio/music is a myth — at least the way it’s defined by the powers that be — and that the future of downloads of any resolution are diminishing. I read an interview with the head of Hi-Res Music at the DEG and the CEO of Astell & Kern on EnjoyTheMusic.com today…it’s the same BS as always.