As we've seen in Part II last week, based on the preference data, there was a pattern for the respondents to this blind test to choose the samples with lower added distortion as sounding "better". I believe this is encouraging for audiophiles who seek "high fidelity" and "accuracy" in the reproduction of music. It's a demonstration that correlates objective levels of distortion with a subjective preference.
Today, as we end off the write-up for this blind test, let's consider the subjective descriptions of what was heard by the respondents. Words describing experience and feelings can be difficult and imprecise, but by correlating how listeners expressed themselves with knowing how they ranked the samples, perhaps we can appreciate the scope of adjectives used when people listen to content with significant harmonic distortion...
I. The "Golden Ears"
To start, why don't we examine what the 5 "Golden Ears" said? These are the individuals who achieved a "perfect" score and were able to list the samples from "least" to "most" distortion and were able to correlate them from "best" to "worst" sounding...
For reference, remember the samples and how much THD applied:
Sample A = -50dB/0.3%
Sample B = -175dB/0.0000002%
Sample C = -75dB/0.02%
Sample D = -30dB/3.2%
Of the group of "Golden Ears", only 2 of them provided a subjective description; remember, this is an international group of listeners so it's not unusual to not receive a subjective comment perhaps due to english as a second language when responding.
Here are the 2 comments:
A sounded slightly more "bright" than B. C sounded very nearly the same as B, very hard to tell the difference. D sounded more "blurry", transients not as well defined.
I heard distortions at the the high and even more at the low range of my speakers.Both of the respondents felt that the differences were only "small". Unlike typical flowery reviews of equipment with all kinds of words about how soundstage varied or depth of perception changed, notice that the descriptions were far from highly detailed for better or worse :-). There are hints of anomalies like tonality difference ("bright"). Resolution loss ("blurry"). And perhaps temporal change ("transient" affected).
II. Those who heard a "HUUUUGE" difference between samplesOkay then, let's consider the folks who felt that there was a "huuuuge" difference between tracks, what did they hear? A sample of some descriptions:
'A' was horrible. I DNF'd on it with Tootsie, and almost DNF'd on Horse. It was like listening to cassette tapes, which I've always hated.
'B' was slightly better than 'D' but maybe almost too much of a good thing. When people talk about clinical, I could see them meaning that in regard to 'B' vs. 'D', esp. with poorly mastered material. 'D' felt ever so slightly smoothed-over. I'm not sure I would pay for 'B' improvement over 'D'.
'C' wasn't overtly objectionable, but compared 'B' and 'D' it was not as good. I would pay to upgrade from 'C'.
Interesting story time: I listened to them in reverse order. So, I listened to the cello piece last. I had pretty much decided on the order by then, but I was reserving judgement on 'B' vs. 'D'. In the cello piece on a couple of his notes (for whatever reason I decided to max vol for the last test) it caused two items touching each other in our living room to resonate. At first I thought it was A's bad recording. Then with 'B' it was worse, almost broken speaker time. 'C' not so bad, by which time I was pretty sure it what was causing it, but I didn't want to get up out of my chair. It came back for 'D', but not as bad as 'B', which "settled it" for me between 'B' & 'D'. Hope that's not cheating. ;)
This was fun. I'm going to have my son re-randomize the files and I'll re-run & re-submit the test with a couple of my headphone setups.Wow. Thanks for the detailed description! Indeed, A, although not the highest level of distortion was the second highest. And I see this respondent got the -175dB sample as "best". This respondent chose B > D > C > A.
Thanks for organizing!
This same respondent also included another comment after listening with headphones:
Hey Arch, I finally re-ran the test with headphones (had my son re-randomize the files for me). It was much, MUCH harder to differentiate with headphones than my loudspeakers from when I tested the first time. With a pair of ATH-M50's I couldn't tell /any/ difference. With the Beyerdynamics DT-1350's, I felt I could tell a difference, but it seemed slight.
As I recall my preferences from the first time using my loudspeakers (best to worst) were: C -- B / D -- A vs. C -- D -- A -- B this time.
Last time I did not like (A) at all. I said I would pay $ to improve my system to have any of the others relative to it.
I preferred (C) a slight amount but not enough to pay for it.
This time it seems I still preferred (C), but not enough to pay for it, and (A) wasn't even last! Wow.
It's interesting (surprising) that I experienced such a big difference between headphones and loudspeakers. I absolutely expected headphones to be more revealing/make me more picky. I guess if one is a glass half-full kind of guy, everything sounds great on headphones! ;~)
Thanks for taking the effort with headphones and commenting on the difference! Not easy at all and I think with different transducers, one could certainly come out with different preferences.
Here's another respondent who thought the samples sounded hugely different:
Here's another respondent who thought the samples sounded hugely different:
I'm not really sure about the rankings. When I used ATH-M40x, I could hear very subtle differences, and it was very hard to tell which was better. Using ER4SR the differences were huge, but still hard to tell which is more distorted. I think C sounded cleaner than the rest, more detailed on the "Horse", with more clear separation of instruments on "Tootie", and more clear sound of background players' breathing on "Clavier"
B sounded noticeably more "mellow", less harsh than the others, with less attack, and much less pronounced voice reverb on "Horse". I think it was maybe the most distorted of the four, as it sounded the most unlike C, but it was more pleasant than A or D due to its mellowness.
A and D sounded dirty, just dirty. Creschendo on "Rhapsody" turned to mess, Drums on "Tootie" sounded "brickwalled", etc.
I think I'd have hard time separating A from D, with C clearly on one side of this spectrum and B on the other distortion-wise.
"Cleaner", more "separation of instruments" used as subjective differentiators. He described the highest THD samples A & D as "dirty, just dirty" sounding. Ultimately he chose C > B > D > A - not bad at all in that he was able to separate the 2 groups: C & B are indeed the least distorted, while D & A were the higher distortion group.
This respondent ranked from "best" to "worst": C > D > B > A. This is an example of one of a number of respondents who felt the -75dB/0.02% sample sounded best. The amplifier used is the Yamaha R-N500 which is certainly not bad at all!Best C has very clear and full sounding harmonics of piano, stings, percussion, voice. All other samples exhibit increasing amount of destruction of these harmonics, not natural and not full sounding.I wonder how much cleaner the clean sample could sound with some equipment upgrades, such as amplifier.
III. Those who felt there was a "big" differenceNext are the folks who felt the differences were still quite audibly significant in a "big" way.
Dear Αρχιμαγο, (since it's Greek), sample B, was clearly the best sounding in my system. You will tell us why, I hope not because of the distortion, hahaha. It became clear to me when I focused on Jenifer's vocals, the center image and placement on the Hootie's song, and very much on the last piano sample. I won't go into analyzing how bad were the worst samples, except saying that sample A was almost annoying.
I rip my cd's, I listen 90% files, I don't stream, and i don't care about the bitrate, I love DSD, BUT It's all in the recording.
I do listen to the same song, in 44 and 192 to determine which is best, and I don't mind admiting the truth when 44 is the winner.
I love great Dynamic Range, and hate highly compressed Hi-Rez files!
I'm reading your pages for many years now, it's been eye-opening and educating for me. I believe that something that measures well, sounds well too, though I hate to admit I've heard great music from guys with valve amps and old speakers.
It's an oxymoron, that although you are a 'Mago, you mostly demystify things in this magical hobby of above all, matching things!Thank you for the nice response my Greek friend! Love the spelling of the name :-). Yes, absolutely, it's much more about the mastering of music than the technical stuff like 44kHz or 192kHz or DSD. Guess what... You're "right"! Sample B is objectively clearly the "best" by a rather wide margin as this was the -175dB/0.0000002% THD track. I see you rated the samples: B > D > C > A.
By the way, you're right that "Archimago" is the name of the conjurer in Spenser's Faerie Queene. In high school, I liked the name and used it as my alias playing D&D. And continued using the name online as my handle over the years :-). These days, for the audiophile consumer, I hope to demystify things (which is not all that magical of course!), but at the same time make it less comfortable for those who sell snake oil and appear to be dealing with magic.
All in all, the destruction of the scene, the hiding of sounds, the loss of overtones and the fullness of music with harmonics.A good description of what high harmonic distortion might do to music. This respondent rated: C > B > D > A. Again, like one of the respondents above, able to prefer the lower distortion tracks over the higher ones.
Bass tends to moveInteresting subjective descriptions. I assume "distortion especially at the top" might be referring to higher frequencies felt to be affected more? This listener rated: C > B > A > D. Again, a general preference for the lower distortion tracks over the higher ones.
Distortion especially at the top
Better balance in the better versions
The worse versions tended to have a bland sound
Track 4 (Lang-Lang): My twelve-year-old plays the piano and rehearses several times a day. Therefore, I got an idea of how the piano sounds when listening to it from different rooms of the house.
Sample A sounds as listening to the piano from the room above the room where the piano is located, doors open (no reverb left)
Sample B sounds as listening to the piano from the room next to it, door closed (sound subdued)
Sample C sounds as listening to the piano from sitting next to it, lid closed (more direct sound)
Sample D sounds as listening to the piano from the room next to it, door open (more indirect sound)
Nevertheless, I always enjoy listen to my daughter playing the piano, no matter in which room I am in!
What amazed me most was that only Sample A of the 4 tracks created a slight stress for my brain in putting a sound image together. For every other sample B-D my brain could obviously refer to a listening experience enabling it to create a realistic sound image.Interesting use of a real-life comparison of how the sound was spatially experienced! This listener selected: C > D > B > A. Another preference for the -75dB/0.02% THD track as "best".
On the other hand, I probably would not buy an amplifier which makes the music sound as it was coming through closed doors…
more detail on question 10: the "Horse" track seemed hardest to hear differences, seemed mostly like a vague sense of "natural presence" vs "distance" (maybe "veiled"). The others varied by what you could hear and for what ranges, but all three seemed easier to assess than that one.
second comment: I think that having the distortion in the same order for all tracks may bias the results. Having first evaluated one set, the listener is then biased to what to expect in the next set. I found myself listening for anticipated changes and seeking reasons to put them in the same order, instead of truly listening to differences. I wonder if you did the same test, but with all sets of tracks randomly sorted, if you would get a different set of answers.Yeah, good point about maintaining the same order on all tracks. Alas, I find it's a balance between simplicity of the test for data collection and also trying to make it not too complicated for listeners, hence the decision to keep the distortion settings in the same order. I think if I were to run this in a lab situation, I would certainly randomize the sequence. You ranked: A > D > B > C. Interesting selection which suggests that you might like the changes in sound harmonic distortion make. Nothing wrong with that of course... You might want to play with Distort more and listen to the effect of harmonic distortion to verify if this is the case.
I find the more distortion there is (in my opinion), the less clearly each musician or singer stand apart from each other, there is more perceived blending (or bleeding) and dynamic range is reduced. There is no place where I could say: Ah, here this sound is distorted (except maybe the fortissimo piano chords of Lang Lang). Good choice of test files with a female singer having some sibilance, a baritone with a grainy voice and slow bowing from Yo Yo Ma also having some grain. As for Lang Lang, he has a hard touch and those block chords stimulate the piano inharmonics a lot, so there is inherently something recalling distortion in all four pieces.
It was not very hard to select the best and worst, but the others were harder to differentiate, so I'm less sure of those.Thanks for the feedback on the music selection and what you heard. Here's what you selected: C > A > B > D. A mixed selection but like a number of others, you also thought the -75dB/0.02% THD tracks sounded "best" and was correct in identifying D as "worst".
I noticed that the higher ranked recordings of D and A had a much broader sense of space and depth that the lower ranked recordings.Interesting. Samples D and A are in fact the ones with more distortion added. You ranked the samples: D > A > C > B. Fascinating - you ranked the samples "correctly" but from highest distortion to lowest! This is actually not unexpected since we are looking at subjective preferences. Increased harmonic content for some people will make the sound "fuller" and this can be interpreted as "space and depth", I suspect.
IV. Those who felt there was "only a small difference"Of all the categories, this is the largest group with 21 individuals. For brevity, I'll select the most detailed descriptions in this group to discuss...
I found that throughout all music samples track B was clearest and most subtle. I would love to know whether B this was the one with the lowest or no distortion added. C has for me the roughest (hoarse sounding) and sometimes blurry sound. Between A and D I could not really make my mind up which one is more natural.Yup, indeed Sample B was the lowest distortion added. And both A and D were the tracks with more distortion added. Your ranking: B > D > A > C.
I think B sounded slightly more transparent, more obvious "air", more detail, less rounded, less full. B seemed leaner than the other tracks. My wild guess is that as you add more and more harmonic distortion the sound becomes fuller, warmer, less transparent. As a result, I am guessing B has the least amount of harmonic distortion.Nice work, another vote for B (-175dB) sounding best. Certainly with less distortion added, there could be the sense of "leanness" - literally less content being added to the signal. Overall, you voted: B > C > D > A.
No matter how hard I tried, I couldn't put my finger on a specific difference between the samples. I certainly couldn't say "I hear distortion". However, there was something said to me "I like this" and "I don't like this" almost immediately (within a few seconds) for each sample. Consistently I did not like Sample C. Consistently I liked Sample A. B and D were somewhere in between, and I would be happy listening to them.
Because I felt that confirmation bias might be creeping in (because I had decided I liked A and didn't like C right from the first track), I asked my son to anonymize the tracks before I tried again using headphones. I still consistently didn't like C (it was bottom or second from bottom across all tracks). For Track A again I picked it as my favourite 3 times out of 4 - but for some reason on Track 2 I placed it in position 3 with D as my favourite.
Complete results in terms of order of preference for each track on headphones:
Track: 1 2 3 4
A: 1 3 1 1
B: 2 2 3 2
C: 3 4 4 3
D: 4 1 2 4Wow! Thanks for the extra work on this and the data! As you've experienced, this is not an easy task and speaks to the subtlety of harmonic distortions when we're listening to actual music rather than just test tones. I see your "final answer" was: A > B > D > C.
B and C are indistinguishable to me, similarly A & D sound about the same.
First pass, C seemed the "cleanest" generally, and I stuck with that initial impression (although I could be wrong: who knows what the actual recordings sound like without running through an amp of some kind?)
The versions I marked low (A&D) sounded a bit clangy in the piano chords and a bit edgy on voice sforzandi (JW's "and the NIGHT...") and a tad less clean.
Clavier revealed almost no differences. I didn't listen to Hootie because I don't like them.
I think choral music and string sections might be more revealing: that always seems to me where systems fall down. Try listening to Hyperion's 40-voice motet "Spem in Alium" on different setups (especially headphones vs speakers+room.)
I'm interested to see the results, as from best to worst, the audible effects to my ancient ears are smaller than I expected.
I might revisit with more revealing headphones (Koss ESP950s)... I doubt that I would hear much difference over speakers.You answer was: C > B > A > D. Great work! Another example where the respondent was able to identify the two low-distortion and two high-distortion tracks as a pair even if not in exact sequence as revealed in the first sentence of this response! Interesting comment about the "clangy" piano chords as a marker of the harmonic distortion.
Re Q11 - I debated with myself how to answer that one. After all, how do I know for sure why I liked one version over another? Maybe I like harmonic distortion, or maybe not. Lets just say I am looking forward to the results. To be clear, I have answered what I believe to be true, but I have not discounted the possibility this belief might be wrong. I could always PX the Devialet for a valve amp I guess :-) Finally, for the record I do have public performance experience, but only as an ex DJ, which doesn't really count for much here ….. Thanks for offering this test, fascinating, and I look forward to being humiliated by the results. Whatever the results, it will be educational, so much appreciated.Final response: D > C > B > A. Not easy! Good discussion on the debate around whether we might like harmonic distortion or not. We'll talk more about this at the end...
In the more distorted versions the bass feel more extended, fuller, some sounds seem to overlap, to mix. Not clear in some examples.
To be honest, I think that a piece or two of full orchestral music (modern soundtrack or classical symphony) and a more modern track (not only acoustical) instrument would have been more representative. The four samples are limited to very few acoustic instruments with a lot of added distortion by its own nature. Maybe, just maybe, more complex tracks or less "acoustics" instruments would have shown clearer differences.You could be right about the tracks selected... Debatable of course since I find large orchestral music also have their own potential confounding factors including the venue of the recording, and the complexity itself might make harmonic distortions less audible compared to a more intimate vocal track or sparse individual instruments where we might be more intimately familiar with what a lone voice or instrument "should" naturally sound like. The response here was D > A > B > C. Interesting since there was a preference towards the tracks with more harmonic distortion, perhaps subjectively it was the other way around? The less distorted tracks were actually the ones this listener experienced to be "more extended" and "full"?
I heard a certain harshness in the track I liked the least and a certain hissing that made the vocal intelligibility worse, but only in direct comparison with the other tracks. This was best heard on my headphones, closely followed by my AV receiver in "Extended Stereo" mode.
It was fun, even though I'm not sure if I actually noticed the harmonic distortions.However, the selected pieces of music do not make harmonic distortion particularly clear. I am curious about the results.Answer: C > D > B > A. You like the -50dB/0.3% track the least. Not sure if this is the reason about the "hissing" but it was certainly one of the more distorted samples.
Traces of scratchiness/edginess on some of the tracks.Answer: D > A > C > B. Interesting! You selected the opposite order in terms of added harmonics! Could the "scratchiness/edginess" actually be things like finger movements on the instruments, subtle vocalizations, or fret noise that were actually obscured by the added distortions?
This was a tough test! The differences were small (even though I suppose the differences in harmonic distortion were large :-) ) and initially I found myself attracted to the "liveliness" of C which I eventually decided was due to more distortion and I changed my ranking. My old ears are not so acute anymore!You said: D > A > C > B! Again, like the respondent above, your selection was in the inverse sequence of the amount of distortion added! Maybe your initial attraction to the "liveliness" of C is actually a reflection that you indeed liked low distortions... But you second-guessed yourself thinking that this was actually due to added harmonics?
This is Taylor C. I had an interesting experience because I forgot that the order of best to worst should end up being the same for all songs, so after listening and ranking my first three songs, I looked back at my notes and realized that I had A and B as the worst two and C and D as the best two. I figured I was developing a preference somehow related to the order of listening because the differences were subtle enough that they very well could have been imagined. But then I re-read your instructions and was pleasantly surprised to see that they were supposed to be like that, which increased my confidence level that I was actually hearing differences rather than imaging them.
Regardless, if I couldn't switch back and forth quickly, I'm not confident I would notice even the highest amount of harmonic distortion you added, which I subjectively interpreted as maybe a very faint buzzing character or less-smooth quality to the tones. Maybe I would notice such a thing for a song I'm very familiar with, but even that is questionable.Hey there Taylor. Yeah, as you can see, not easy eh? Even with the ability to switch quickly between samples... How much more difficult it would be if one had to switch cables or move components in and out of the system! I see you ranked: C > D > A > B.
V. Those who felt there was "very little difference" - basically very subtle...
Next, here are some comments from those who felt the differences were minute and "not worth an upgrade" if this is the difference between products.
At first I listened with my speakers to all the tracks straight through, ranking each one from best to worst. It was hard noticing any difference, but I was pretty sure all the A samples were the worst (most distortion). As for the other ones, it was really all over the place.
Then I listened with my headphones. This time I listened in reverse order – from D to A for each set, instead of A to D. It was actually harder to tell the difference and I was no longer sure that A as the worst.
I finally put every set of samples into audacity (I DID NOT look at the spectrogram, I promise) just so I can jump between them quickly and to a/b test between different samples (using headphones and speakers). This was the least revealing test, and even jumping quickly back and forth while looping A second of audio yielded no noticeable difference.
I tried also listening at different volumes. It didn't help.
My ranking from best to worst is pretty much a guess, except from sample A which I ranked the worst, which was the most consistent thing that I could hear (though not in the quick a/b testing). but the difference could be imagined for all I know. I chose the piano peace as the most revealing, though this is also just speculation and could be an imagined effect.
Subjectively, this is all music I never listen to normally (I'm a proud metalhead). I wouldn't know how it supposed to sound or what am I supposed to listen for when I listen to it. The only thing close to "audiophile" experience I had with this music, is the piano piece had the most realistic imaging (funny thing – when I put on the sample of the rhapsody with the headphones the first time, at first I thought I was still hearing it from the speakers). I think this is just due to it being a single instrument recorded in a way that realistically captures the acoustic space around it. Still pretty boring stuff, though ;-)Thanks for taking the time Mr. Metalhead. Clearly not the "usual" type of music for you :-). Your final ranking: B > C > D > A. Congratulations! Despite all the years of head banging and listening to loud (distorted) music, you did a great job with selecting the least distorted tracks as "best"! Keep rockin'...
Если честно, то заметил я лишь (по моему убеждению) максимальный уровень искажений и только на одном жанре, с насыщенной басовой партией (Tootie by Hootie), все остальные практически невозможно было ранжировать - расставил практически интуитивно.
Translation: "To be honest, I noticed only (in my opinion) the maximum level of distortion and in only one genre, with a rich bass part (Tootie by Hootie), it was almost impossible to rank all the others - I arranged it almost intuitively."Nice, a response written in Russian (I believe the survey system tagged this as a response from Ukraine). As I mentioned previously, this is an international effort :-). The respondent answered: B > A > C > D. Exactly as the comment suggested. He correctly picked the "best" as lowest distortion Sample B, and the "worst" as highest distortion Sample D. Difficult to arrange the middle samples. Well done!
Horses: The "Sss" seems cleanest a I sorted, however the difference is not reliably detectable (ABX compare) Ranking: Best: D, B, C ,A: Worst
Rhapsody: the ranking is different again not reliably detectable: Best: C, B, A, D: Worst. In rhapsody the parts with octave jumps and thins sound is the best for me to test.The ABX Test failed in all 4 pieces. I could not reliably detect in the blind test the ranking (:
This means the distortion is not at all obvious.
In a personal experiment I detected that with music rather high distortion is not detectable in the amount of 3-5%. With a sinewave, the distortion was detectable at an amount of around 0.5%. The sine sounded harsher with distortion.
Conclusion: In music the masking effect is strong.Thanks for commenting on your ABX results! Your final ranking was: D > B > C > A. It is interesting that your ranking for "Rhapsody" would have been closer to the sequence from low to high distortion added. Thanks for also commenting on the personal experiment and your result with a pure sine wave being significantly easier to hear harmonic distortions by the time 0.5% THD added. I have done some tests myself and indeed my result is similar; depending on specific amount for each harmonic and higher harmonics also less easily masked by the fundamental.
Room characteristics and noise floor will be just as important as how much is spent on equipment. And the type of equipment. My MC225 probably adds more distortion that what was added to the samples.
Big fan of your blog, been a reader for many years.Thanks for the feedback man. Yup, indeed all kinds of factors will play a part in the final sound / preference. Indeed, your McIntosh MC225 tube amp will have its own unique harmonic signature which could be quite high, potentially further modified by the type and age of the tubes used. I see your final selection was: A > D > B > C.
First of all, I really tried to find any difference as that seemed to be very hard. After a while I became convinced that track A was sounding warmer, while D was more "airy", less "filled" with sound.Cool descriptions of the 2 tracks with the most amount of distortion added. I see you selected: A > C > B > D. Yup, the "airy" track is the one with the most distortion as you suspected in the ranking.
The ordering I performed above is totally bogus. For Clavier, I felt track A was "best", and for Rhapsody it was track D. For the other 2 pieces I could not rank them. I might have heard slight differences in "transparency", "fog", or "artificiality", but on different listenings these would be apparent on different tracks - no consistency. So, for the most part I heard no rankable differences.
I will admit that I have little patience for A-B type tests. I do much better hearing differences in long-term listening. The other day I was playing an album I am very familiar with - the MQA version on Tidal. After about 5 cuts I realized that it just wasn't moving me the way I expected, the enjoyment was lacking. I switched to the same album in my local library and immediately noticed that it was more alive, had more transparency, and better dynamics. I thoroughly enjoyed the second half. (I am not blaming MQA here, but rather the different mastering employed. Not massive compression/equalization, but enough to rob the album of some of its vibrancy.)
Anyway, thanks for creating and administering this test. I am looking forward to the results and your analysis of them. Pardon me but now I have to go listen to Famous Blue Raincoat.Thanks for the "bogus" rankings man :-). I see you selected A > D > C > B. Possibly a preference for the sound of the samples with increased harmonic distortion? Totally agree with your comment on mastering of course. Perhaps the Tidal MQA version used a more compressed master - I have seen this happen over the years with the MQA version being slightly louder, more fatiguing. If this hypothesis true, I bet that if you could quickly A/B compare the same song from your local library and the Tidal MQA version, you would easily hear the difference without needing any kind of long-term listening and also be able to decide which one sounds "better" as well.
Hope you got a chance to enjoy Famous Blue Raincoat again!
VI. In conclusion...As I mentioned, I didn't include all comments I received above; the ones I did include I believe represented the general flavour from my data set. This should give the reader a glimpse into the different ways that listeners subjectively described the same set of songs and sample variations.
I think this brief comment from a listener who heard only a "small difference" sums things up quite succinctly:
Harmonic Distortion seems not to have a big influence on sound quality - to my surprise!There were a few "surprised" comments like this that the listener was unable to hear much of a difference despite having high expectations. This is why I believe it's important for hobbyists to try listening for themselves instead of just reading (and potentially repeating) what others claim. We see comments about "harmonics" all the time in magazines and among online posts. I suspect if any one of us were told that our amplifier had "3% harmonic distortion", we'd be rather aghast at the terrible performance! But how many of us have taken the time to measure our own equipment to appreciate the level of nonlinear distortion present or listen with this knowledge for ourselves to experience what the distortions actually sound like?
Although I received the results from 67 listeners, I know many more have downloaded the test tracks and likely have listened as well but not filled out the survey. I'll leave the download as linked in the Blind Test Invite up as long as I can. By all means, use it like the demo tracks I made for jitter and for the "bits are bits" discussion. As shown in a few responses above, it's quite possible that some listeners preferred the sonic effects of the nonlinear distortion added. On the other hand, others assumed that characteristics like "fullness" or "air" might represent the effect of these distortions but in fact were found in the tracks with lower distortion once we "unblinded" the selection. Such is subjectivity and speaks to the limits of the human perceptual and cognitive systems.
At the end of the day, I think we can see from the subjective descriptions that there really isn't some common language shared in these responses. Nor is there a universal type of preference for each of us. Despite this, as shown last week, when we look at the average responses as a "group of audiophiles", we can see that there is still a preference for the "low distortion" samples B (-175dB/0.0000002%) and C (-75dB/0.02%) over the "high distortion" samples A (-50dB/0.3%) and D (-30dB/3%). This is the "power" of increasing sample size and blind testing when teasing out subtle effects.
Remember to consider these facts when we see testimonies from self-proclaimed "Golden Ears" and individuals who might be outspoken about what they believe they hear. Unless one knows that the individual has had experience as a "trained listener", keenly disciplined and aware of what distortions actually sound like, we can never be sure that what sounds "good" for any one individual actually translates to a broad group of audiophiles. For all we know, the individual could have a preference towards the sound of higher distortion!
What I can say from the "objective" results based on this blind test is that as a group, listeners still prefer the sound of lower non-linear distortion, and one can certainly determine whether audio components add distortion through objective measurements. I believe this knowledge is of benefit to the "more objective", and rational audiophiles.
Again, thank you to the 67 respondents - see, no need to fear blind testing! :-)
And of course a special thanks to Paul K for Distort and the thoughtful collaboration over the last few months!
As the world turns, often in very unpredictable ways, I wish everyone health, safety, and peace - while enjoying music along the way as we head into summer.