I agree on everything he's saying, especially when it comes to blind listening tests. Eliminating as much (confirmation) bias as possible is the only way to objectively chose the best sounding piece of gear or combination thereof.
Cable rolling is quite the pointless waste of time and money once you get past the obvious basic issues like noise rejection and shielding. Everything else is snake oil. I walk away when people start talking about the sound of cables. Its 99.9% BS.
The thing I disagree about is the sound of speakers and how specific measurements interpolate to a given audible trait or signature. Speakers can be a weird animal to judge. So many tiny things will affect the sound of a speaker, especially in a given room or placement. The issue with relying on measurements is repeatability. Without it, you can't say specific parameters are the cause or symptom of what you're hearing.
There are also huge neurological differences between people's hearing ability and general psychology. Hearing is a highly complex human sense and it varies significantly from person to person. There are some people in the audio profession who will harp on specific generalizations of how some measured technical specifications relate or equate to what soecifically we hear. There are way too many minute little combinations of measured (and undetermined / overseen) parameters relating directly or indirectly to what we can exactly perceive with our ears.
If it were possible to isolate specific indivdual (or groups of) directly measured data sets, then reliably trace this directly and repeatably back to what we hear via blind testing, you could safely state it were possible to measure and isolate the data of exactly what we hear. This is unfortunately not possible to the extent that is required to make these claims. There are still certain things we hear that cannot be traced or linked specifically to certain exact measurements sets of data at this point in time with our current technological abilities. We're getting closer to having this ability, but for now its not completely possible.
We know that some types of measurements and their derived data correlate to bad sounding audio. However, much of the individual data can't be specifically isolated to be the cause of some exact issue we can hear, let alone place an individual label on. Only some combinations of data can be linked to specific audible traits and occurrences.
This reverse process shows the ability of the human ear to figure out certain specific issues without the need for test equipment. We can't however quantify this in even close to the precision and exact numbers as calibrated test equipment can, given once we have determined the possible cause.
So what I'm saying in a nut shell is that we can measure certain parameters much more accurately than our ears will ever be able to discern. We can't however exclusively rely on test equipment to fully isolate a given abnormality, issue or trait we can perceive with our hearing. The threshold for detection can be much lower detecting some types of distortion using test equipment vs our ears. The test equipment can however not quantify how the measured distortion will affect or impact our hearing, regarding how fatiguing or enjoyable it is to listen to the speaker under test. So basically, test equipment is excellent at giving objective and repeatable data, but the data itself can't be used to correlate how it will impact our ears, specifically with the combinations of some distortion types and the proportions of individual harmonics.
Sorry for the long post.
Cable rolling is quite the pointless waste of time and money once you get past the obvious basic issues like noise rejection and shielding. Everything else is snake oil. I walk away when people start talking about the sound of cables. Its 99.9% BS.
The thing I disagree about is the sound of speakers and how specific measurements interpolate to a given audible trait or signature. Speakers can be a weird animal to judge. So many tiny things will affect the sound of a speaker, especially in a given room or placement. The issue with relying on measurements is repeatability. Without it, you can't say specific parameters are the cause or symptom of what you're hearing.
There are also huge neurological differences between people's hearing ability and general psychology. Hearing is a highly complex human sense and it varies significantly from person to person. There are some people in the audio profession who will harp on specific generalizations of how some measured technical specifications relate or equate to what soecifically we hear. There are way too many minute little combinations of measured (and undetermined / overseen) parameters relating directly or indirectly to what we can exactly perceive with our ears.
If it were possible to isolate specific indivdual (or groups of) directly measured data sets, then reliably trace this directly and repeatably back to what we hear via blind testing, you could safely state it were possible to measure and isolate the data of exactly what we hear. This is unfortunately not possible to the extent that is required to make these claims. There are still certain things we hear that cannot be traced or linked specifically to certain exact measurements sets of data at this point in time with our current technological abilities. We're getting closer to having this ability, but for now its not completely possible.
We know that some types of measurements and their derived data correlate to bad sounding audio. However, much of the individual data can't be specifically isolated to be the cause of some exact issue we can hear, let alone place an individual label on. Only some combinations of data can be linked to specific audible traits and occurrences.
This reverse process shows the ability of the human ear to figure out certain specific issues without the need for test equipment. We can't however quantify this in even close to the precision and exact numbers as calibrated test equipment can, given once we have determined the possible cause.
So what I'm saying in a nut shell is that we can measure certain parameters much more accurately than our ears will ever be able to discern. We can't however exclusively rely on test equipment to fully isolate a given abnormality, issue or trait we can perceive with our hearing. The threshold for detection can be much lower detecting some types of distortion using test equipment vs our ears. The test equipment can however not quantify how the measured distortion will affect or impact our hearing, regarding how fatiguing or enjoyable it is to listen to the speaker under test. So basically, test equipment is excellent at giving objective and repeatable data, but the data itself can't be used to correlate how it will impact our ears, specifically with the combinations of some distortion types and the proportions of individual harmonics.
Sorry for the long post.
People's hearing does vary, but I do think it's on a scale of very good (no damage) to very bad (lots of damage). There are a small percentage of people who can hear 21kHz or above when they are young. They've usually lost that by their 30s, and since most high end system owners will be over 30 and have accumulated hearing damage over the years from going to clubs, live concerts, transport noise, power tools etc I think we can say hearing and the threshold of hearing is more or less a known quantity.
With speakers I think he was saying there is still some confirmation bias to deal with but there are also obvious sound differences between speakers so bias is less of an issue with speakers, though it still exists.
With speakers I think he was saying there is still some confirmation bias to deal with but there are also obvious sound differences between speakers so bias is less of an issue with speakers, though it still exists.
There is a lot of visual bias in judging speakers. However, having been a judge in a large speaker contest, speakers will surprise you! A lot of them don’t sound anything like you would expect, both for better and for worse.
I’ve attended well organized speaker contests where the judges could not see the speakers. This seemed to me a necessary part of the judging - until I became a judge in another contest. Every speaker I saw generated some sort of expectation, most of my expectations were proved wrong. 😉
The problem with testing in general and controlled bind testing in particular, since audiophiles seem to be under the impression that’s the gold standard, is when results are negative or inconclusive they do not have much meaning since so many things can and do go wrong with the test or test system. The only path to find out the truth is conducting many tests, with different systems and different listeners and let the accumulation of results be your guide. A single has limited meaning.
Last edited:
It is, in science. It's not just limited to audio. Double blind tests are a good standard. And if they are negative that does tell you something.
The only way to fairly judge speakers is through blind listening tests. Playback volume has to be the exact same as well as music. The setting its performed in should be visually neutral, without visual distractions or any other sensual input ie. artwork on walls, food odors, soft chairs, etc. We also used to do 2 sets of calibrated levels to determine low level detail retrieval and dynamic capabilities, using different music for either level sets.
A pre-recorded spoken dialog between a man and woman was also played back at actual real world speaking levels. This was very revealing of issues some speakers exhibited in the midrange. Many people were surprised how much easier they could judge the midrange resolution using actual human speech. This was Harbeth's specific scheme in developing their own speaker designs, which made alot of sense. You then realize most people can judge speech accuracy even though the test subjects didn't have an accurate perception of what music was supposed to sound like.
A pre-recorded spoken dialog between a man and woman was also played back at actual real world speaking levels. This was very revealing of issues some speakers exhibited in the midrange. Many people were surprised how much easier they could judge the midrange resolution using actual human speech. This was Harbeth's specific scheme in developing their own speaker designs, which made alot of sense. You then realize most people can judge speech accuracy even though the test subjects didn't have an accurate perception of what music was supposed to sound like.