Music on the Brain

Our latest poll question—Do You Prefer the Sound of Digital or Analog Audio Media?—has inspired more comments than any I've posted up to now, and I'm grateful to everyone who has added their two cents to the discussion so far, as well as those who will do so in the future. This is exactly what I had hoped these questions would stimulate—a lively but respectful discussion of the issues that concern all who enjoy the audio/video hobby.

I'm a bit surprised that digital has a strong lead over analog, especially given that many commentators express a preference for the sound of analog while acknowledging that digital is far more convenient. So I want to make sure everyone understands that the question asks whether you prefer the sound of digital or analog audio media, not which one is better in one way or another or which one you listen to more often. In that respect, there can be no objective answer as C seems to want; after all, preference is entirely subjective.

Clearly, C is a devout materialist—the only reality is that which can be measured objectively, at least in principle. I agree that scientific measurement is critically important; in fact, I'm trained as a scientist, and I rely on objective measurements quite heavily. But in my belief system, this is not—nor can it ever be—the entire story.

For example, I do not believe that all human experience can be completely explained by biochemical processes as C maintains. Consciousness itself cannot be fully explained in this way, since the relationship between mind and body is a two-way street with each having the ability to influence the other. The placebo studies that C cites prove this by unequivocally demonstrating the power of the mind to affect the body.

C's proposed double-blind listening test is an excellent idea, and I hope someone performs it someday. (To recap, using the highest-quality equipment, record the same performance on both analog and digital media, then master each one to vinyl and digital, so you end up with four recordings—AA, AD, DD, and DA. Play all four to a large sample of listeners in a double-blind process and record their responses.) I suspect the results would resemble similar tests performed with different grades of audio cables—experienced listeners would be able to discern the differences between the various recordings more reliably than naive listeners.

Also, C seems to imply that if the human psyche is real, it will be captured in a recording. Quoting him (assuming C is male) and editing for clarity, "If someone consistently preferred the AA and DA recordings, maybe it is just the vinyl noise and distortion they hear and unconsciously associate with the positive 'human psyche' feel. Then it has nothing to do with digital being unable to capture this mysterious 'human psyche' because the original master of the DA vinyl is a digital recording; one cannot introduce 'human psyche' after the recording has been completed."

I did not say that recording in analog or digital "captured" the human psyche; I said that many believe that analog recordings reach the human psyche more deeply than digital recordings. "Psyche" is just another word for consciousness, mind, or soul, the existence of which is often denied by devout materialists. But even though the psyche/mind/soul is indeed mysterious—i.e., not well-understood—that does not mean it doesn't exist. I believe it does exist and can be influenced by music, though not in a way that can be easily measured.

I'll even go one step further—I believe the psyche/mind/soul can be influenced differently depending on the medium, and there are even objective measurements to support this. For example, I've heard of brain studies indicating that highly compressed digital audio affects the limbic system—the part of the brain involved in emotion—less profoundly than uncompressed digital audio. (I've been trying to contact the researchers who conducted these studies, but so far, no luck.) Similarly, I'm not surprised to hear many people say that analog audio on vinyl is more engaging than digital audio.

As I said earlier, I'm trained as a scientist, and I place great stock in objective measurements and testing. However, I've learned that measurements alone aren't enough—for example, reviewers have often observed that amps and speakers can measure well but sound bad and vice versa. Thus, I agree with brad.clarkson that experience generally trumps tech specs in the end. And personal preference trumps everything else, so listen to as many different devices and types of recordings as possible and buy what you enjoy the most. If you do, you can't go wrong.

Share | |
C's picture

Thank you for expanding on this immensely stimulating discussion. I obviously read a lot of your contribution to UltimateAVMag, which I appreciate and admire your expertise knowledge and thoughts in this wonderful passion we all share. I do like to take this opportunity to make some corrections of my last comments on the previous blog discussing preferences over analogue or digital audio, while also present some more questions on the present topic of

Barry Willis's picture

Asking "Analog or digital?" is always a great conversation starter, but one that's ultimately about as unproductive as a religious discussion. Both analog and digital technologies have their strengths and their flaws. Both can be used to render recordings that are "accurate" (in the engineering sense) and emotionally satisfying. And both can be abused to create sonic crap. Cases in point: cheapo cassette tapes and low-rez MP3s.

Hard-core analog folk can go on and on about warmth and emotional honesty vs. the cold alienating sound of digital, but ask them about any of analog's obvious flaws -- ticks and pops, surface noise, speed fluctuations, inner-groove distortion, pre- and post-echo -- and they are likely to say "That doesn't bother me." If you dig a little deeper, they may admit that all those flaws are actually sort of reassuring, a sonic safety net.

Digital has huge advantages over analog in terms of ease of editing and archiving, and in terms of noise-free copying. But there are artifacts that can make digital recordings sound cold, alienating, or unnatural. Neither technology is perfect.

The best recordings combine proven techniques from both camps. A current audiophile fave is Shelby Lynne's "Just a Little Lovin'," her tribute to Dusty Springfield. It was recorded in real time to 2" analog tape, which was used to make a digital master. The CD sounds excellent -- warm and enveloping, without a hint of digital harshness. Analog and digital are both wonderful technologies if used with judicious good taste.

The comments from "C" are interesting for their technological orthodoxy. His proposal for an exhaustive comparative study of mixes of analog and digital is a good theoretical exercise but is so riddled with practical problems that it would be impossible to implement.

C's picture

First of, I admire your writings such as the "The Art of The Scientific Illusion" that appeared in Stereophile many years ago... But it always puzzled me when one points out that certain proposals for certain studies are riddled with problems, yet failed to identify the problems with any specificity, then immediately concludes that the proposed test is "impossible to implement" as a generalized conclusion... If that is the attitude going around, NONE of any of the research we see today will be performed. In research, there are always problems, which further requires the exact identification of "problems"...

In practicality, the testing area can be of a stage with a narrow set of seating area to prevent severe changes in auditory perception in the test subjects, let us say 5 at a time, 3 together in a row, and 2 immediately behind the first row. The stage will be prepared with the live instruments, and the playback transducers (speakers). The analogue and digital recording and playback equipment will be wired in the back stage.

Musicians and singers will rehearse the exact musical pieces on stage in the same physical spots, with average volume measured. Then the proposed recording is to be recorded under the same physical setting to produce the 2 versions of masters: digital and analogue. The masters will be shipped off to manufacture the playback copies in best vinyl and digital formats (such as PCM with high sampling rate at 96khz or above and bit depth of 24 bits). Once these AD, AA, DA, DD playback copies are finished, the tests shall begin.

During the actual test, the performers will be placed in the exact spots where they were when the recordings are made. Another rehearsal will take place and measurements made to ensure the live event volume is identical to the recordings. Of course the live version will be musically slightly different from the recorded versions, but this is the bench mark. Furthermore, it would be interesting to playback a blank vinyl in the background while some live events are going on or combined with a DD recording to further confuse the subjects, in order to reduce psychological bias. All subjects who had their hearing tested will be futher instructed to sleep at least 8 hours the night prior to any testing (those with minor deficits are to be INCLUDED in the test for the purpose to study confounding variables) and required to fill out a questionnaire to ensure that they are feeling their normalized physiological state before the start of the test. Half of the study groups will be exposed to the live event in random order while being BLIND folded during the entire test. The other half of the group will be exposed to the live event first for each set of tests, with full visual clues and told that the live event is the benchmark, THEN blind folded to hear only the recordings following the live event . For all tests, the musicians are on stage through out the entire test, and get a signal from testers when to perform and when just to sit there. If they need to cough during the recording portion of the test, even better to confuse the subjects. Scoring for the subjects will be performed by pre testing instructions to understand and familiarize with the actual scoring system, and deliver the scores in the forms of clicking a remote switch (like those voting system seen on TV games) so the effects of peer bias is reduced.

Finally, the use of brain imagery or blood lab work is another study all together, but I only threw that in for the purpose to demonstrate the vast sources of confounding variables and measurement techniques available..

Many tests, done not nearly as precisely or elaborately as above, have been done many, many times in the past (and many times under live situations) mainly for testing speaker performance. Subjects are usually recruited in the academic setting. I am sure there are tons of both younger and older university engineering, medical, psychology, and music students and faculty members who will participate with the above experiment (that is how a lot of research gets their subjects). Impossible to conduct? Ken Pohlmann from Sound & Vision is a professor at University of Miami... (hint hint) This can be a big task to take on, but a study that is tedious is very different in meaning from being "impossible"...

I am sure many also rushed to generalize that tests like these were impractical when the idea was merely suggested, and then immediately declared the idea for testing is "so riddled with practical problems that it would be impossible to implement" without any meaningful details or critical investigation. Attitudes such as this will just end any quest to learn new information, and at the same time preserving and propagating the "status quo" legacy of presumptions and myths that is factually suspicious without the support of refined scientific evidence. If that is the case, why bother testing, or even bringing any discussions to light, if we are all just to generalize and stick to untested presumptions and myths? We would still be so backwards and be playing with Edison's gramophone if the continual quest for for new information that is supported by scientific evidents did not exit... Lastly, sound and music reproduction, just like any type of human engineering such as airplanes, is entirely based on and made possible and advanced by science, not by mere believes and faith like religion. And the last time I read UltimateAVMag, this is a technical publication discussing the engineering products of music reproduction, not a music magazine... Shouldn't any productive conversation about a product of science be solely based on scientific facts?

Jarod's picture

All things considered I prefer analog recordings to digital. On a proper set-up analog is my heaven. Being a guitar and drum player and a lover of music in general listening to an LP really puts me right there in the same room with the performer.

C's picture

For the test that I proposed, I can already see a few confounding variables that need to be addressed...

I just realized that a lot of people do not even know what the real instruments sound like anymore since 99% of their exposure to musical sounds are via recordings or artificially amplified at a live events. Furthermore, human hearing is notorious for poor psychoacoustic memory, even for musicians who play the instruments all their lives. People often are very sure of what they "think" a certain live instruments "should" sound like... There needs to a golden standard benchmark: live acoustic event, and a control group of subjects who are not exposed to the golden benchmark at all. Therefore, the 4 versions of the recordings (AA, AD, DA, DD) need to be combined with the live instruments recorded by the same performers at the same environment, side by side, played back immediately and randomly for comparison. Score sheets should include 3 sets of correlating variables: subjective, objective, and proper identification. Examples are: dynamics: which one has the best perceived dynamics, which one the subject "preferred", and which one was live, AD, AA, DA, or DD. The list can go on: noise, detectable forms of distortions, tonal balance (frequency response), height, depth, size, and location imagining, etc. I know many of these live vs. recorded-on-the-spot demonstrations have been performed by a few loudspeaker companies at trade shows (which used to be open to public...) such as Duntech... However, I do no think the live source was compared to various formats of the recordings as I mentioned above...

Furthermore, the recording and playback devices, especially the transducers that convert acoustic energy to electrical signals or vice versa, such as the microphones, mic preamps, and speakers should be calibrated to as linear as possible so no equipment in the chain will benefit by certain minor inaccuracies from these devices. Do the mic and preamp and the speakers have inaccuracies such as slightly raised or depressed curves at certain frequency bands, and how all this behaves with variating amplitudes and impedance of the rest of the equipment in the chain, digital or analogue, which can make some recording formats sound more real, or in some cases, more preferable to the subjects despite being inaccurate?

Subjects of both the control and experimental groups should be of a large random sample (>30), should also all be tested for conductive and sensory hearing impairments. Older aged subjects who tested positive for the inability to detect sound pressure waves above 8Khz at a certain threshold amplitude should have their perception of tonal balance critically studied for other possible interferences with other tested variables... Other data such as musical and recording instrument training, age, sex, should also be considered... Control group will not be exposed to random exposure of the live sound stimuli and their results to be compared with the experimental group.

The combined effect of tonic and phasic neurosensory adaptations involved in hearing should also be taken into consideration in this test, which means duration and breaks between testing materials have to be carefully arranged so there still exists an immediate A/B comparison to avoid lags in human hearing memory, but also to avoid fatiguing and receptor adaptation. Then there is the types of musical and instrument selections, simple, complex, etc, and also should include various types of human voices.

The subjects of both groups should be exposed to the acoustic stimuli in sets according to type of acoustic event. Such as a drum set, male speaking voice, female singing voice, chorus voice, a string quartet, piano solo, harp solo, etc. Each set should contain at least 8 samples: Live, AD, AA, DA, DD, plus 3 more random formats of the 5 versions of the acoustic event to confuse the subjects further to prevent possible bias (control group will obviously have 4 more random formats).

While one is at it, why not take PET Scans or real time MRI as well...?

C's picture

After much thought, the experiment that could not be specifically sited but casually mentioned brings about even more questions...

From the way this test was described, it is stating that uncompressed music elicits a response in the human limbic system. This seem to suggest a direct cause and effect relationship: uncompressed music causes a response in the human emotional center. The interpretation of some can assume that uncompressed recording "reaches the human psyche" in a mysterious and undefined manner somehow...

First of all, is the compressed version of recording perceptually identical to the uncompressed version to the test subjects? And was this inability to distinguish either version of recording pre-tested in a scientific manner with a meaningful 75% or greater statistical incidences? If the compressed recording is so much worse in quality that the subjects can tell that it is an inferior recording, then the validity of this experiment is, frankly, 100% garbage. If the test subjects can perceptually distinguish which recording is compressed or uncompressed, then there is bias. One might as well only compare uncompressed recordings against another set of the same uncompressed recordings, except process one set with severe distortions such as clipping in the analogue pre-amps to make it perceptually sounding worse for the test... I assure you that the group exposed to the distorted recording will possibly have their limbic system light up all over the place as well, but of negative emotions, even that they were exposed to only uncompressed recordings and compressed versions were all omitted from the test... In that case, the test is ONLY about the effects of perceptually higher quality sounding recording on the limbic system regardless of being compressed or not.

Furthermore, was there a control group that did not get exposed to uncompressed recordings? OR to reverse the thesis of this experiment to prove that "compressed recordings do not elicit responses from the limbic system", take out the compressed recordings for the control group. Will there still be variations in the limbic system response?

What about the various external and internal confounding variables such as if the test subject was too hot, hungry, tired, getting inpatient, started to dislike the testing process, etc?

And how does limbic system lighting up automatically mean a positive or negative feeling in the subject? That is to presume that the limbic system is indeed responding to perceptually identical but different quality recordings?

And how do we know the result is not showing some correlational effect of the interactions of all the confounding variables? How were the testers sure that this indeed was a cause and effect relationship? Did they follow up with more studies to alter the proposed cause and look at the effects/results?

Is this a single case study or a large randomized sampling of subjects with a control group?

We all know that sometimes a very poorly designed experiment is actually better not done at all, because it can perpetuate false information. If the reality is actually A, and the test biased all the subjects to choose B, then people actually believe false B as the reality, and when questioned, confidently brings out the poorly designed experiment as valid scientific evidence... That is actually more harm done...

Enter your Sound & Vision username.
Enter the password that accompanies your username.