mikeaj
Headphoneus Supremus
- Joined
- Jan 10, 2010
- Posts
- 1,639
- Likes
- 110
Sorry, I missed this stuff before.
Quote:
Quote:
---- First part is response to top section ----
Based on just the measurement data, which is done with equipment with very high accuracy, I agree it's a stretch to conclude that this accounts for everything the ear and brain hears or perceives. However, what I'd say you're ignoring is the large body of evidence supporting the claim that extremely small differences in sound are impossible for humans to perceive. There are many studies employing controlled listening, where people were in fact not able to discern very small differences; sometimes the differences are those that many audiophiles say should be apparent. I think it's also common sense and intuitive for most people, that two objects machined to be the same size might look so similar that people couldn't tell them apart by looking, that two batches of food processed by the same factory might be so similar as to be indistinguishable by taste (yet they may appear to taste significantly differently based on your mood, what you previously just ate, and what aspects you were focusing on), and so on. The same should hold true for two audio samples as well, as demonstrated by all those previous studies.
You don't need to account for every single factor to make claims that are most probably correct, based on the current understanding of electrical theory, psychoacoustics, and so on.
I think there are two key unintuitive notions to overcome:
(1) that two significantly different electronic circuit designs (audio or not) can have extremely similar performance characteristics in certain relevant ways -- I think a stronger background in EE would help here
(2) that our senses and perceptions, which serve us so well in so many circumstances, can be unreliable -- I think being a superhuman would help here
The question is then about whether or not the O2 or any other amps are at the level where they're sufficiently close to the ideal. Personally, I'm not going to say I'm 100% certain that the O2 is at this level, because I'm no expert in pscychoacoustics, so I'm just parroting the 0.005% and whatever thresholds that seem to have been established. And even if those have been established by previous studies, maybe they're wrong. Also, I think it's possible but very unlikely (based on an interpretation of systems theory), that an amp could perform as well as the O2 with all the measurements V checked, yet somehow behave significantly worse with some other kind of audio input and headphone load. The more linear an amp is--which the low distortion figures point to--the more it should probably behave like an ideal LTI system and thus have a very predictable and close to ideal input/output relationship, no matter what the input.
Again, this is a real-world system, so performance is going to depend slightly on the temperature, the electricity from the wall, the positions of every individual electron in the room (which are unknown obviously), and so on. Nothing behaves exactly like the textbook theory, but is it close enough?
As far as I can tell, your conclusion that the O2 does not achieve "wire with gain" performance, is based on your own listening, maybe the listening of others, and maybe some other factors. Correct me if I'm wrong. As far as reputable academic publications are concerned, this is not very convincing evidence. Furthermore, it's less convincing IMHO than the kinds of evidence and theories I and others have been pointing towards. Again, if listening conditions are not controlled very carefully, even well-trained listeners can exhibit consistently biased tendencies. As one quick example:
http://seanolive.blogspot.com/2009/04/dishonesty-of-sighted-audio-product.html
---- Now this is the response to the two bottom paragraphs posted ---
Regarding V's listening test, he's already said that he (and others) did it. I wouldn't take negative blind listening results too seriously from somebody not expecting to hear differences, anyway. For all you know, he could be intentionally sandbagging.
As for Benchmark's side, it would be interesting to see their response. I wouldn't be confident in a 100% honest and straightforward answer though, as they have products to sell. That said, much of the cost of the DAC1 goes to profits, R&D costs, the DAC and all of its functionality, and so on. Furthermore, regarding the DAC1 HO, it is co-located next to noisy digital electronics, which gives it a disadvantage compared to the O2. Furthermore, the output power levels into lower impedances seem to be higher than the O2 can deliver. Also, the output impedance seems to be much lower, though arguably it seems like the O2's is already low enough? These extra abilities and features above the O2, require a more complicated or expensive design.
The O2's core design is stupidly simple and seems to represent pretty much the best you can do with that kind of effort and cheap parts. It's pretty much the simplest thing you can do that's not a CMoy. The achievement of the O2 is in recognizing that stupidly simple designs can have good performance, putting in the hard work in terms of the PCB layout and components to optimize the design, choosing the correct cheap parts instead of the most fashionable ones that may perform worse, and doing everything with the constraints that it must run off of batteries and use all through-hole components.
P.S. For the record, I think some arguments being made on both sides are overstepping their bounds and need to be reworked.
Based on just the measurement data, which is done with equipment with very high accuracy, I agree it's a stretch to conclude that this accounts for everything the ear and brain hears or perceives. However, what I'd say you're ignoring is the large body of evidence supporting the claim that extremely small differences in sound are impossible for humans to perceive. There are many studies employing controlled listening, where people were in fact not able to discern very small differences; sometimes the differences are those that many audiophiles say should be apparent. I think it's also common sense and intuitive for most people, that two objects machined to be the same size might look so similar that people couldn't tell them apart by looking, that two batches of food processed by the same factory might be so similar as to be indistinguishable by taste (yet they may appear to taste significantly differently based on your mood, what you previously just ate, and what aspects you were focusing on), and so on. The same should hold true for two audio samples as well, as demonstrated by all those previous studies.
You don't need to account for every single factor to make claims that are most probably correct, based on the current understanding of electrical theory, psychoacoustics, and so on.
I think there are two key unintuitive notions to overcome:
(1) that two significantly different electronic circuit designs (audio or not) can have extremely similar performance characteristics in certain relevant ways -- I think a stronger background in EE would help here
(2) that our senses and perceptions, which serve us so well in so many circumstances, can be unreliable -- I think being a superhuman would help here
The question is then about whether or not the O2 or any other amps are at the level where they're sufficiently close to the ideal. Personally, I'm not going to say I'm 100% certain that the O2 is at this level, because I'm no expert in pscychoacoustics, so I'm just parroting the 0.005% and whatever thresholds that seem to have been established. And even if those have been established by previous studies, maybe they're wrong. Also, I think it's possible but very unlikely (based on an interpretation of systems theory), that an amp could perform as well as the O2 with all the measurements V checked, yet somehow behave significantly worse with some other kind of audio input and headphone load. The more linear an amp is--which the low distortion figures point to--the more it should probably behave like an ideal LTI system and thus have a very predictable and close to ideal input/output relationship, no matter what the input.
Again, this is a real-world system, so performance is going to depend slightly on the temperature, the electricity from the wall, the positions of every individual electron in the room (which are unknown obviously), and so on. Nothing behaves exactly like the textbook theory, but is it close enough?
As far as I can tell, your conclusion that the O2 does not achieve "wire with gain" performance, is based on your own listening, maybe the listening of others, and maybe some other factors. Correct me if I'm wrong. As far as reputable academic publications are concerned, this is not very convincing evidence. Furthermore, it's less convincing IMHO than the kinds of evidence and theories I and others have been pointing towards. Again, if listening conditions are not controlled very carefully, even well-trained listeners can exhibit consistently biased tendencies. As one quick example:
http://seanolive.blogspot.com/2009/04/dishonesty-of-sighted-audio-product.html
---- Now this is the response to the two bottom paragraphs posted ---
Regarding V's listening test, he's already said that he (and others) did it. I wouldn't take negative blind listening results too seriously from somebody not expecting to hear differences, anyway. For all you know, he could be intentionally sandbagging.
As for Benchmark's side, it would be interesting to see their response. I wouldn't be confident in a 100% honest and straightforward answer though, as they have products to sell. That said, much of the cost of the DAC1 goes to profits, R&D costs, the DAC and all of its functionality, and so on. Furthermore, regarding the DAC1 HO, it is co-located next to noisy digital electronics, which gives it a disadvantage compared to the O2. Furthermore, the output power levels into lower impedances seem to be higher than the O2 can deliver. Also, the output impedance seems to be much lower, though arguably it seems like the O2's is already low enough? These extra abilities and features above the O2, require a more complicated or expensive design.
The O2's core design is stupidly simple and seems to represent pretty much the best you can do with that kind of effort and cheap parts. It's pretty much the simplest thing you can do that's not a CMoy. The achievement of the O2 is in recognizing that stupidly simple designs can have good performance, putting in the hard work in terms of the PCB layout and components to optimize the design, choosing the correct cheap parts instead of the most fashionable ones that may perform worse, and doing everything with the constraints that it must run off of batteries and use all through-hole components.
P.S. For the record, I think some arguments being made on both sides are overstepping their bounds and need to be reworked.
