Several minutes between listening to samples? The problem with that is that auditory memory is actually very short. With an A/B/X comparator, the listener can switch as often or as rarely as they please. But any wait time between the samples dramatically reduces the validity of the comparison. If that was an important part of their protocol, it increased, rather than decreased, the subjectivity of their tests.
The brain shouldn’t “rest” when comparing two things. Imagine how poorly a text subject would do on one of those “find the difference between two pictures” tests if they looked at one picture for 200 seconds, then had it taken away, allowed their brain to “rest” for a few minutes, then was allowed to look at the second picture. Seriously, how do you think they’d do? Why should the less reliable auditory memory do better than visual memory?
FWIW, I read in a newspaper that audiophiles were spending a lot of money buying the original Playstaion console because it apparently could play CDs with better sound reproduction than a high end stereo system. It had some sort of special processor that produced a very clear sound signal or something.
From “Inaudible High-Frequency Sounds Affect Brain Activity: Hypersonic Effect” by Oohashi and al., Journal of Neurophysiology 83:3548-3558, 2000.
They did three series of tests: one where they looked at EEG response, one where they took a PET scan and another where participants wrote down their subjective assessment of the sound. All three showed significant response to full-spectrum signals.
The ABX test will tell you if you can consciously perceive a difference between two sources. The hypersound research shows that even though you cannot tell two sources apart in such a setting, people will nevertheless tend to subjectively prefer the source with high frequency components. Specifically, they rated these sounds as “softer”, more “reverberant”, more “balanced”, more “comfortable” and “richer” than they audible-frequency-only signal.
I think this may in part explain why some people can claim to prefer high-quality recordings while at the same time being unable to identify them in ABX testing. It’s likely not all self-deception, as is often claimed. As a matter of fact, this was part of Oohashi’s motivation:
I searched and found it. Their methodology is a mess. If I read this correctly, there were two different sets of speakers!
LFCs and HFCs were separately amplified with P-800 and P-300L power amplifiers (Accuphase, Yokohama, Japan), respectively, and presented through a speaker system consisting of twin cone-type woofers and a horn-type tweeter for the LFCs and a dome-type super tweeter with a diamond diaphragm for the HFCs.
As two devices can’t occupy the same space, they had a different position relative to the listener. And any possibility of a double-blind test is eliminated when you hear two different sounds sources. And the methodology is pretty haphazard:
The level of the presented sound pressure was individually adjusted so that each subject felt comfortable; thus the maximum level was approximately 80-90 dB sound pressure level (SPL) at the listening position.
There’s not a word mentioned about matching the two sound sources within .1 dB like the study of audibility of high-bitrate improvements by David Moran and Brad Meyer linked to by Mix magazine. Tragically, the original paper is only available for download if you pay $20 ($5 for AES members). But here is the abstract:
Claims both published and anecdotal are regularly made for audibly superior sound quality for two-channel audio encoded with longer word lengths and/or at higher sampling rates than the 16-bit/44.1-kHz CD standard. The authors report on a series of double-blind tests comparing the analog output of high-resolution players playing high-resolution recordings with the same signal passed through a 16-bit/44.1-kHz “bottleneck.” The tests were conducted for over a year using different systems and a variety of subjects. The systems included expensive professional monitors and one high-end system with electrostatic loudspeakers and expensive components and cables. The subjects included professional recording engineers, students in a university recording program, and dedicated audiophiles. 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. The noise of the CD-quality loop was audible only at very elevated levels.
Why? Presumably the subjects couldn’t consciously hear the high frequency content. The experimenters played the music with and without the inaudible content, and both times they heard the same audible sounds through the same speakers.
Even if hypersonic frequencies really did make a difference in the sound quality (and this is still a pretty big if), it makes no difference since next to nobody owns speakers that can play back those frequencies. As gaffa points out, in the study they had to use different tweeters to play back the higher frequency part.
It would also require that all the recording equipment preserved those frequencies as well (I wonder if the Moran & Meyer study accounted for this). It has essentially no bearing on the LP/CD issue. I’m also not sure why my post was quoted, since I was referring to material that couldn’t possibly have higher frequency components in the song data, and mentioning my ability to distinguish them anyway (in response to the claim about how high a bitrate was needed).
The way I read it, it’s a four-way speaker with a crossover. Although a custom job, it’s not an unusual listening setup. Two woofers, two tweeters, with one tweeter only getting HFC. I think you will find it very difficult to get a 100kHz flat response with a single driver.
I don’t get your objection to the double-blindedness of the study. The piece was played in full-spectrum and audible-spectrum in random order. Since the HFC cannot be consciously perceived, there was no way for either tester or subject to known which was being played at any given time. There was either sound coming out of all four drivers, or only the three drivers assigned to the LFC.
I’ll grant you that I would have liked to see a bit more discussion of measured sound levels between the samples. However, in their setup, the audible-frequency signal is the same in both LFC and LFC+HFC samples. Since HFC are imperceptible, the apparent sound level between the two samples ought to be the same.
The claims of hypersound research don’t contradict those of Moran and Meyer, because they deal with conscious discrimination. Oohashi were interested in unconscious effects.
I simply quoted you because you mentioned ABX.
They used a special system to get a flat response to 100kHz. “Regular” speakers do indeed reproduce sounds above 20kHz, it’s just that the frequency response isn’t likely flat in that range anymore.
LPs cannot reproduce high frequency signals. Like I wrote, the size of the needle’s tip, for one, limits the frequencies that can be accurately reproduced. However, a turntable will likely produce some high frequency content, nevertheless, in the form of noise and distortion. I am wondering (not claiming) if, considering the hypersonic effect, could this high-frequency energy account for the so-called “warmth” of analog sound?
I would be interested in seeing the results of an experiment where the hypersonic component wasn’t taken from the recording, but consisted entirely of noise.
But can inaudible, subconscious hypersonic sound that seems to be at best discriminated by presence really be counted towards sound quality? Wouldn’t that be like saying that it’s more pleasant to listen to music in a somewhat warm room than at sub-zero temperatures, and then claim that therefore temperature affects sound quality?
Perhaps one should do a double-blind study where one group gets the ‘full spectrum’ signal with the hypersonic part, another only gets the audible part, and a third group that gets the audible part plus random hypersonic noise/a constant/modulated hypersonic signal; if there’s no difference between groups one and three, I’d say that the hypersonic is more an environmental effect than part of sound quality.
I feel bad knowing that I won’t live to see the last whining, smelly, vinyl-loving hippy drop dead so that this non-argument can finally be ignored.
It isn’t that vinyl sounds better, its merely that vinyl sounds different!! We’re not talking about acoustical engineering here, but the emotional attachment of memories. Once there’s no one left who remembers how the suck-ass, oops, I mean ‘original’ vinyl recordings sounded no one will care.
As for modern vinyl releases, niche marketing for club DJs and fanboys, nothing more.
I’m not just bitter, I’ve been an audiophile most of my adult life. When CDs arrived it was a religious experience because it was so freaking obvious to me that vinyl recordings were not ‘warm’. Vinyl recordings were, ahem, shit. Dragging a diamond across a piece of plastic? Oh yeah, that’s a great sound capturing medium!
I’d say yes. Simply put, if you play two recordings to a group of people, in random order, and ask them to rate on a scale how “soft” or “hard” it sounds, and you find a significant statistical trend, then I’d say that counts towards sound quality. Interestingly, while there was a quite significant trend for qualities like “soft” and “hard”, there wasn’t for “like” and “dislike”. So, de gustibus and all that…
I don’t know, I feel that’s cheating a bit – let’s say you’d discover some subconscious signal that gives people who listen to it a nice warm and prickly feeling, and you subsequently encode that into your audio recordings. Have you then raised their quality?
I feel strangely the same about my old VHS tapes, especially the ones I taped off TV, with some of the films followed by something like an episode of Mission Impossible. DVDs just seem to encourage interupted viewing. Sit down, watch a favourite scene, then remember something else more important to do instead.
Dad thought something similar might work with chemical film photographs and digital photographs. His idea (which I think the boards debunked) was that you were looking at something with a resolution in the range of atoms compared to one that was so many dpi. But there’s a limit to the resolution an eye can discern anyway, so it in the end it makes little difference.
Is there such a discernible difference according to the human ear?
I only scanned the article, but I didn’t see any reference to a double-blind testing protocol. If it’s mentioned, could you please quote the relevant section?
I didn’t see anywhere that they went to any effort to match the sample levels at all. This looks like a parody from the Journal of Irreproducible Results. It has the form of science without actually containing any science.
The Ethan Winer piece is very relevant here. If they heard one piece, then left the seating location to have a PET scan, then sat back down to hear another piece, they are very unlikely to be in exactly the same location. Different location, different sonic conditions. The chart in the Winer article pointed at 3 to 6 dB difference at different parts of the spectrum with a 4" difference in location. Hell, they may have been measuring the effects higher blood pressure on hearing caused by sticking the subject’s head into a PET scanner!
Did you also play the CD in direct comparison in the same environment? I have done so with the LP vs CD in a high-end soundroom, and I felt the CD was noticeably superior, specifically regarding the very low frequency “heartbeat” in the song. “Time” is actually what I use to test potential speakers, there is such range in that song that any speaker or amplifier limitations become immediately apparent
Yes, the Moran & Meyer study used equipment that could reproduce extended frequency range. The source was a SA-CD player capable of reproducing up to 90k and electrostatic speakers.
The signal from the SA-CD was split and one output was run to the A side of the A/B/X comparator. The other output was run into the input of a HBB CD recorder set in record monitor mode, and the output of that went to the B side of the A/B/X comparator. The listener would push the button to switch to a randomly selected input, either the direct, very high quality and extended-range SA-CD or the 44.1 k, 16-bit bog-standard CD.
Only a small percentage of the listeners were able to do better than random chance.
I’m sorry, but reading your objections, I think you should have read the article closer. First, there were four distinct experiments, two with EEG, one with a PET scan and a psychological evaluation test.
EXPERIMENT 1. To examine the physiological effect of sounds with an inaudible frequency range, 11 subjects were presented with the FRS, HCS, and baseline conditions.
(…)
In accordance with conventional recordings of background EEG activity, subjects were asked to keep their eyes naturally closed during the experiment to eliminate any effects of visual input. The presentation of the sounds in both FRS and HCS conditions lasted 200 s, which included the entire piece of music. The baseline condition also lasted 200 s without sound presentation. The inter-session intervals were 10 s. Two recording sessions were repeated for each condition in the following order: baseline–FRS–HCS–FRS–HCS– baseline.
EXPERIMENT 2. The validity of the digital audio format internationally employed for CDs was evaluated under more ordinary listening conditions. Seventeen subjects were presented with sounds using a cutoff frequency of 22 kHz, which corresponds to the upper range of sounds recorded by a CD. Subjects were then asked to keep their eyes naturally open as they usually do when they listen to music. The open-eye condition was also appropriate to control the subjects’ vigilance. Each subject was presented with four types of conditions: FRS, HCS, and baseline, as in Experiment 1, plus LCS to elucidate the effect of an HFC when it is presented alone. As in Experiment 1, each condition lasted 200 s. Before the actual recording sessions, HCS was presented once to familiarize the subjects with the experimental environment. To avoid any influence by the order of presentation, the four different conditions were performed in random order across the subjects. After a 10-min rest, the same four conditions were repeated in reverse order. Neither the subjects nor the experimenters knew which conditions were being performed.
PET measurement and analysis
The sound presentation equipment was installed and calibrated in the PET laboratory of Kyoto University Hospital. Subjects lay supine, with their eyes naturally open, on the PET scanner bed in a quiet, dimly lit room. Their heads were fixed in individually molded helmetshaped rests that were contoured to leave their ears undisturbed. The distance from the speakers to the subjects’ ears was approximately 1.5 m. As in the EEG study, special attention was paid to the immediate environment to minimize the subjects’ discomfort. Six of the subjects were studied using FRS, HCS, and baseline conditions, and the other six were studied using FRS, LCS, and baseline conditions. The order of the conditions was randomized across the subjects and a total of six scans was performed on each subject with intervals of 7 min.
Psychological evaluation of sound quality
We also evaluated the subjective perception of sound quality. Since the subjective impression of sounds is closely related to the subjects’ psychological condition, this evaluation was performed separately from the EEG and PET experiments. We used the same piece of gamelan music as was used for the EEG and PET experiments. First, a pair of FRS and HCS, each lasting 200 s, was presented. The order of the conditions was randomized across the subjects. After an intermission of 3 min, another pair of FRS and HCS was presented in reverse order. Therefore the stimuli were presented in an A-B-B-A fashion, in which FRS and HCS were assigned to A and B or B and A, respectively, in a randomly counterbalanced way across the subjects. Neither the subjects nor the experimenter knew what the sound conditions were, although they did know that the presentation was in an A-B-B-A fashion. The subjects filled out a questionnaire to rate the sound quality in terms of 10 elements, each expressed in a pair of contrasting Japanese words (e.g., soft vs. hard). Each element of each condition was graded on a scale of 5 to 1. The scores were statistically evaluated by the paired comparison method described by Scheffe´ (1952).
In all but the first experiment the order was randomized. All four experiments were performed separately. No one was moving from one room to another to have a scan taken. Also, notice how in the EEG and PET test they were also taking baseline measurements, which is why you can rule out the effect of the environment on the listener.
Yes, head position has a significant influence on how a sound is heard. And, yes, with the exception of the PET experiment the authors could have provided more detailed information about relative position to the speaker. However, even if listeners were moving around, considering that the order was randomized, how do you explain correlation to show up in all tests for the group?
