Perfect Sound Eventually? MP3s, CDs, and High Res Audio

“Every disease is a musical problem; every cure is a musical solution.” ― Novalis

When CD players first hit the market, some people, including many audiophiles, complained that they sounded like crap. I lived through this period and for me early CD playback was painfully bad—brittle, shrill, dry, and lifeless. Of course other people argued, and some still do!, that this perception was nonsense, CD-quality being, for all intents and purposes, perfect in theory. Much better than analog. They were of course wrong in practice.

We have since learned that there were in fact unwanted negative sonic artifacts generated by early CD players including significant jitter that could cause high-frequency distortion and crappy anti-aliasing filters (see What are Digital Filters and Why Are They Required In Today's Audio DACS? by Resonessence Labs Technical Staff for more great information on filters) which introduced ringing artifacts into the audible spectrum. We're talking about plain new fashioned bad sound.

The same story goes for early MP3s. Musical bits that were deemed to be perceptually unimportant by the numbers and in listening tests were in fact missed by some listeners, including many audiophiles. If you recall early marketing for the MP3, the tech was sold as being "CD-quality". Hardly. Even the notion that early lossy encoding was audibly transparent is disputed by people like Chris "Monty" Montgomery who claims to have been able to tell the difference between different encoders, "There was a time in the 1990s when I could identify every major mp3 encoder by sound (back when they were all pretty bad), and could demonstrate this reliably in double-blind testing".

CD replay has improved substantially over time as has lossy encoding. With the ever increasing popularity of computer audio, file-based playback, and CD-quality streaming, we are inarguably in a very good musical place. With improved jitter performance, better filters, the ability to play from memory, and more, I would suggest that CD-quality replay can sound simply stunning.

“The question is, are we done? Have we reached the point where we have 'perfect sound forever' from 16-bit/44.1kHz files?”

The question is, are we done? Have we reached the point where we have "perfect sound forever" from 16-bit/44.1kHz files? Some people think so, others do not. The dividing line between these two beliefs is not drawn by the typical objectivist v. subjectivist split. Rather there are people arguing "the science" on both sides.

"Monty" Montgomery suggests that 16/44.1 is all we need:

"This means we can use low rate 44.1kHz or 48kHz audio with all the fidelity benefits of 192kHz or higher sampling (smooth frequency response, low aliasing) and none of the drawbacks (ultrasonics that cause intermodulation distortion, wasted space)." And "16 bits is enough to store all we can hear, and will be enough forever."
Monty proponent Justin Colletti from Trust Me, I'm A Scientist (facetious title)" recognizes that higher sample rates can in fact sound better:
"When properly designed, a slightly higher sample rate may allow us to smooth out our super-high frequency filters and keep them from introducing audible rolloff or ringing which may be perceived by younger listeners (if they’re paying any attention.)

...

"There are definitely some converters that sound significantly better at a higher sampling rate than at a lower one, even in a blind test. But strictly speaking, the problem isn’t with the lower sampling rate – it’s with the converter."

This last point is certainly an important one. When we listen to different file resolutions on our DACs, are we really testing the file resolution or are we testing our D/A converter? I think the answer in this limited case is the latter, we are testing our D/A converters. In order to properly suss out whether or not higher sample rates and greater bit depths do in fact sound better than CD-quality, we'd have to run exhaustive tests on lots of DACs. I'm not of aware of anyone who has performed such tests.

The oft referred to paper by Meyer and Moran titled Audibility of a CD-Standard A/DA/A Loop Inserted into High-Resolution Audio Playback is brought up to prove that people cannot distinguish between CD-resolution and higher rates. The actual tests performed compared SACD and DVD-Audio to CD and they concluded, "The test results show that the CD-quality A/D/A loop was undetectable at normal-to-loud listening levels, by any of the subjects, on any of the playback systems." Since that time, a number of people have criticized these results. For Mark Waldrep of iTrax, his main criticism is that the "high-res" recordings used for the tests were not high-res recordings but were sourced from analog tape. Others, including John Atkinson of Stereophile, have called the paper "methodologically flawed".

In a paper titled "The Audibility of Typical Digital Audio Filters in a High-Fidelity Playback System" authored by Helen M. Jackson, Michael D. Capp, and Robert J. Stuart of Meridian Audio, their "double-blind psychophysical test" concluded:

"Results suggest that listeners are sensitive to the small signal alterations introduced by these filters and quantization. Two main conclusions are offered: first, there exist audible signals that cannot be encoded transparently by a standard CD; and second, an audio chain used for such experiments must be capable of high-fidelity reproduction."
Of course there are people who criticize this test as well. One such person is Robert Adams of Analog Devices. As the Boston Audio Society reported, Adams "pointed out that [sic] the filters used in the testing were much sharper than those used in typical pro audio convertors. In fact the total impulse response length of their (Stuart's) digital filter is in the range of 8ms. A typical converter filter is in the range of .5ms to 1ms. So this is overstated by a factor of at least 8."

Dan Lavry of Lavry Engineering has also argued the benefits of higher sample rates in his paper "The Optimal Sample Rate for Quality Audio":

"Although 60 KHz would be closer to the ideal; given the existing standards, 88.2 KHz and 96 KHz are closest to the optimal sample rate."
Here's John Sau of Benchmark Media from his post titled "High-Resolution Audio - Sample Rate":
"Jumping to an 88.2 kHz or 96 kHz (2X) sample rate should eliminate all concerns about exceeding the limits of our hearing."

So many points of view, so few acceptable tests, so much science, so little agreement. What's an audiophile to do?

Nothing's Perfect But It Doesn't Have To Be
Where does that leave us? If you want to snuggle up with some of the most current science, go with 24-bit/96kHz. Of course that does not mean that all 24/96 files are created equal. The obvious caveats being that the quality of the original recording matters most, which has been the case since the dawn of recorded music, as does the provenance of the recording in question. Some people, like Mark Waldrep of AIX, feel that an analog recording does not benefit in any way from being put into a 24/96 or higher "bucket". Others like HDtracks, Qobuz, Acoustic Sounds, Pono Music, etc. feel differently. Who's right?

You are. If something sounds better to you, it is better. If CD-quality sounds best, fine. If 320kbps lossy files float your boat, cool. If 24/96 files sound best, there's no reason to buy higher rates. If 24/192 or DSD are your musical ideal, wonderful. You know what works best for you and your DAC. Personally, I enjoy CD-quality, higher resolutions even when they originate from analog tape, DSD, FM radio in my car, and vinyl. I've also had the pleasure of listening to CD-quality and higher resolutions through over 100 DACs in my system and lots more at Hi-Fi shows, dealer events, etc. and I've found there are consistently real sonic benefits to be had from resolutions greater than CD-quality. YMMV, of course.

While I hate to sound like a broken record, listening to music on the hi-fi is not a listening test nor is it a science experiment governed by theories despite anything anyone tells you. What listening to music on the hi-fi is governed by is enjoyment. As we've seen with early MP3s and CD replay, our perception identified real sonic problems before the science set about fixing them.

COMMENTS
CG's picture

Nah...

As somebody once explained to me about a different hobby, "This is not life or death. To a lot of people it's far more important than that!"

Michael Lavorgna's picture
...once and for all in a relevant scientific manner is to arm wrestle.
Alex Halberstadt's picture

Well written and thought-provoking, Michael. Some years back, I conducted a highly unscientific experiment where I had a friend randomly play various resolutions and formats of the same track (encoded from the same CD rip) while I listened on a pair of good headphones. Eventually, I realized I *could* differentiate them—not by their sound, but by how well they kept my attention. An 192 kbit/s MP3 made me want to stop listening almost immediately; a CD-quality AIFF made want to keep listening and often allowed me to connect with the music emotionally. Eventually, I found I could even distinguish AIFF and ALAC versions of the same track apart with something like 80% accuracy. I didn't have higher-resolution files on hand, but what this experience suggested to me is that the way we hear and respond to music is highly complex, more so than the existing measurements tell us. Keep up the good work!

Michael Lavorgna's picture
...but by how well they kept my attention."

Exactly Alex. Excellent point.

Cheers & thanks.

DH's picture

I have some amazing sounding CDs - if they all sounded like that I probably wouldn't be interested in hi-res. Of course, maybe those same albums would sound better in hi-res, but I don't have any way of knowing.
My experience has been exactly what you said - if the original recording and mastering is well done, it will sound great in any format, even mp3. I can enjoy mp3.

I do agree with Alex, though. On the really good recordings you find yourself doing more foot-tapping, and notice all of a sudden that you are especially enjoying yourself.

Michael Lavorgna's picture
...I find that lossy formats not only do not hold my attention, I find them irritating over time. I just lose interest in listening. The lower the bit rate, the faster I lose interest.

In terms of recording quality, this issue is endemic to recording regardless of format. I've heard great sounding LPs and not as great sounding LPs. Does that mean I'm going to stop buying LPs? Or only buy great sounding LPs? Nope. It means that some records I buy will sound better than others. Of greater importance is my interest in the music.

In the end, I try to buy the best sounding version of the music I'm interested in. With LPs, I've found that typically means buying as close to the first pressing as possible. With digital, I've found that high-res typically sounds better than the CD equivalent.

Alex Halberstadt's picture

"I just lose interest in listening. The lower the bit rate, the faster I lose interest."

That sums it up for me, too. And as we lower the resolution, digital files seem to have the uncanny tendency to subtract engagement without altering the sound.

I think the distinction you're making between sound and engagement is one reason the Pono has been received by some members of the tech press with such vitriol. What Neil Young promised was to restore some of the emotional connection with the music. He did not promise better-than-ever sound that would make people cross-eyed with excitement during blind listening tests.

I wish we had a better vocabulary for talking about a technology's ability to engage us with the act of listening—which has nothing to do with bass, treble, harmonic distortion, or transient response.

Michael Lavorgna's picture
...for talking about a technology's ability to engage us with the act of listening..."

I do too.

VK's picture

One thing that i find relevant to the discussion, is that "a theory" is something completely different from "absolute truth" or "absolute knowledge". And for what i see in various sites and forums, people simply ingnore that, or at least, are not aware of that. Some audio theories exist for years and decades, and as you pointed out Michael, there's no agreement! We will always be learning, and in the meantime, stop the small talk to go to the point of all this: to enjoy music!!! Some appears to have forgotten that this is the most important thing!

Best regards!

Michael Lavorgna's picture
...in the meantime, stop the small talk to go to the point of all this: to enjoy music!!!
I would just add - !.

Cheers.

Reed's picture

I'm really enjoying my CD player right now. I tried the server/high rez route for several years. I found the high rez stuff unnaturally clean sounding for my taste. My current CD Player cost me $350 on sale (normally $499) and is unbelievably good. It is by far not the weak link in my system right now. I wish CD playback could have been this good from the start.

Michael Lavorgna's picture
...but I'm happy to hear you're enjoying your CD player.

Cheers.

Dr. AIX's picture

I held a session here at the AIX Records studios over the weekend for a select but small group of audiophiles. I played some of my 5.1 96 kHz/24-bit surround music tracks and the group was thoroughly impressed. If any of your readers want to hear original 96/24 PCM music files produced without any EQ or dynamic processing, just click on the FREE music link on the Real HD-Audio blog site and I'll send you a link to the FTP site. Cheers,

Rick Tomaszewicz's picture

...performance quality is often unrelated to recording quality or resolution, the cost of high res downloads dissuades many of us from exploring new artists. High res streaming, with access to millions of tracks, can solve this problem.

Alternately, thrift store hunting invites horizon-widening based on what's available. (I just picked up a couple of ECM recordings at Sally Anne. Never heard of the artists, but trusted Manfred Eicher, the producer and label owner. )

Once in a while, just as you happen to be there, staff puts out a box of CD's or vinyls which just arrived at the thrift store. You start pawing through the stuff and, lo and behold their previous owner had magnificent taste and treated his music with kid gloves! Now you're looking at a curated collection! I've often gone home with a box of more that 50 albums for $50. Bliss for a month.

rexp's picture

Redbook digital recordings (on vinyl) have been acceptable to most audiophiles since the 80's, the problem has been until recently Redbook playback has sucked big time and very few have heard how good it can sound as most CD/file players still suck.

PDQ.Bach's picture

…if we understand science properly: as an endeavour to learn and to understand.
What many misunderstand as "theory" is just dogma, or misapplied preconception, unsupported by real evidence.
'Data' is not the plural of 'anecdote': this works both ways.

Beyond a love of music and many youthful years spent recording, in the field and in studios, I developed an acute scientific interest in the matter when my father became ill with Parkinson's Disease.

A palliative treatment for PD en vogue some years ago was music therapy. It turned out that music from early CD players or the earliest MP3 made my father's symptoms worse. The very same music from LP on a decent turntable was acceptable. I did a series of tests together with his neurologist: playing my original studio tapes via the same monitors, level-matched etc., but in different formats (tape, DAT, cassette, LP, CD-rip, MP3, AIFF file). Obviously the source players differed according to format.

Tape was soothing; DAT and LP neutral; cassette, CD-rip and MP3 worsened the patient's symptoms, in that increasing order. AIFF, when played via one of RME's early ADC/DAC units, was only slightly worse than tape (=soothing). The effect was more pronounced when I copied the master tape again to AIFF, this time at 24bit/88.2kHz. The best effect was achieved with AIFF at 24bit/88.2kHz played via Quad II amps and Quad electrostats. This had a really calming effect on my father's symptoms, reproducibly so, despite the measurable harmonic distortions.

Regarding the Meyer and Moran AES paper, I think some of the criticism was misdirected. Acoustically, they knew what they were doing (and they had a professional lifetime to show for it). Their methodology was flawed, but in more subtle ways than most of their critics were able to judge. The experiment design was ill-suited to the task, and the statistical analysis suffered from the lack of an appropriate, explicit hypothesis to be tested, instead of the conventional Null Hypothesis Significance Testing that still prevails in many such experiments. A real pity, in view of the sample size and number of trials. If the detailed original data were made available, I wonder whether a re-analysis wouldn't yield more satisfactory results.

Michael Lavorgna's picture
Thanks for sharing.
ktracho's picture

I remember reading back when 300 DPI laser printers were just becoming available for purchase at a reasonable price (under $1000) that increasing the resolution to 600 DPI would make the dots small enough to be at the limit of human ability to distinguish individual dots. (Consequently, I printed my thesis using larger fonts, and then reduced it when I photocopied it to 100% cotton paper.) However, increasing resolution further to 1200 DPI made text less tiring to read, and 2400 DPI just looked sharper still, even though the human eye cannot see additional detail in the text. (Just compare the readability of something you print on your computer with that of a professionally printed book, one that wasn't produced using laser printer, that is.) Very few people argue that there is a benefit as far as readability is concerned to making higher resolution displays than what is available in current phones/tablets. What would happen if someone did try to argue that we need higher resolution displays for reading text?

ezrazmus's picture

it's only with audio that people argue against higher digital resolution, and i think it's partly because of the horrible image of the audiophile.

a rich nerd with a ponytail talking about vinyl will never lead a cultural trend.

no one screams "snake oil!" when a new printer comes out with a higher resolution.

no one screams "snake oil!" when a new camera comes out with a higher resolution.

no one screams "snake oil!" when a new TV or computer monitor comes out with a higher resolution.

but if it's audio, for some reason, the internet idiots come out and proclaim -- BS! buy new headphones and enjoy your mp3's!

it's a special group of idiots - these types, because they masquerade as intelligent.

ezrazmus's picture

Awesome article, I agree with you. The actual science is clearly on the side of a higher resolution than a 16-bit data space can handle. 18bits would be better, 20 is enough for almost anything in PCM stereo that current microphones can capture and speakers can reproduce.

I don't think there's anything wrong with 24/96 as the new standard but I do hear 'more' at 24/192 when rendered on the ponoplayer. It's really about the air and the timing at that level.

If you hear the room at 24/96 and you can place the instruments in their basic location, at 24/192 you start to enter the room and the instruments start to envelope around you.

Layers are revealed like an onion - the bass strings, the synth patch's squawk, more guitar distortion, a string part, the slight creak on the floor tom, but especially the mix part of the mix. Does that make sense?

The art of recorded music is in the delays, pan, movement, and space in the room, and that opens up even more at 24/192 - assuming you can render it properly.

X