If there is some "weird corner case" it will show up in a difference or least mean squares difference, unless one is arguing for parapsychology.
-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Master Thread: Are Measurements Everything or Nothing?
- Thread starter amirm
- Start date
If there is some "weird corner case" it will show up in a difference or least mean squares difference, unless one is arguing for parapsychology.
Ours was in a rack with a 20" fan on "high" bolted to the back of the rack, with space above and below the amp. The switch on the fan was hardwired, and the power switch to the amp supplied the power. Couldn't turn on the amp without the fan. This might be getting off this master thread topic though.
jareinertson
Active Member
- Joined
- Jul 4, 2025
- Messages
- 114
- Likes
- 87
MattHooper
Grand Contributor
- Joined
- Jan 27, 2019
- Messages
- 11,126
- Likes
- 19,766
I noticed there were some back-and-forth about the legitimacy of hearing differences between some digital sources.
For me, digital is a solved problem, I use a benchmark DAC, and I don’t think about ”sonic differences between CD players/DACs.”
However, as I’ve mentioned before, I had some experience in the late 90s perceiving what seemed to be some obvious differences between some CDPs and a DAC. I’d been able to borrow various gear from friends or dealers and at one point I had the Meridian 508.20 CDP, a Museatex Bidat DAC (which had a volume control), and the good quality Sony CDP my wife and I had been using up to that point.
I was surprised at what seemed to be very distinct Sonic differences between each of them. But since it was general in my position, I shouldn’t hear a differences between competently designed digital gear, and understanding the possibility of bias effects, I did a blind test and I was able to easily identify each of them.
To get some criticism and feedback I presented the feedback and results online to a place where a number of engineers and “objectivists” hung out. Anybody here who goes back far enough might recognize names like Arny Kruger, Howard Festler, Stewart Pinkerton, and I believe our own J.J. was there as well.
I received some excellent feedback and constructive criticism as to how I could tighten up the test (which included switching from matching volume level using a RadioShack metre, to using a voltmeter to measure the speaker terminals to match output).
I redid my tests and presented those results again to the gang.
I’m just going to copy and paste my old post describing my attempts at the blind test and the results.
It might be mildly entertaining for some to read about an audiophile attempting blind testing at home, and also why it’s not exactly an easy thing to pull off. And of course I leave opinions of the test results up to the reader:
—————————-
—————————-
I wrote the original post entitled "My BLIND TESTof CD Players -
here’s the results…."
That post detailed my experience in hearing differences between my
Museatex DAC and my Meridian 508.20 in a blindtest that I
constructed. Many helpful types pointed out to me that the Radio
Shack meter I used was not the proper device to employ if I wanted to
match output levels with real accuracy. "Get thee a voltmeter and do
it right" they said. And so I went forth, and lo, I did find a
voltmeter.
I borrowed a voltmeter from the tech department at work (I work at a
large sound editing/mixing facility).
The sound technicians showed me how to work it, and best deploy it,
for my test.
I had been working up a sweat imagining having to beg my spouse to
help me in another audio-nerd test. In a stroke of luck, my
father-in-law happened to be around to help me do these tests. He’s a
classical music nut and owns a beautiful-sounding music system. He is
also a hard-nosed engineer, and does not bother listening to
components that should sound the same (re: amps, CD players etc.) -
his late 80’s CD player and amp are doing just fine, thank you.
Luckily he’s more sympathetic to the cause of good sound, and I was
able to squeeze many more test scenarios out of this session than in
the first one.
Right then. The test.
I chose a comfortable volume at which to listen, and used the
voltmeter to measure the level at the speaker terminals. My speakers
measured a few tenths-of-a-volt different in level, so I used my
pre-amp’s balance control slightly to even them out. I set both the
Meridian and the Museatex DAC to 1.72 volts. As in the previous test,
they measured the same, and I could not perceive a volume difference
between the two units.
The method we used was the same as the last test (as this was deemed
acceptable by the critics). He did not know which DAC was connected
to which inputs. I told him not to simply switch between the two
sources constantly, but to make the switching unpredictable and
‘random.’ A sans music pre-test showed that, with a few ‘fake’ switch
movements of the source selector in between source selections, I could
not reliably guess which source had been selected. I was blind-folded
and we did not talk to each other during the test, except when I said
"switch." My FIL (father-in-law), kept track of my correct and
incorrect answers.
Here’s the score for TEST A:
Meridian vs Museatex
Out of 18 trials
Incorrect Guesses: 1
Correct Guesses: 17
For some reason I guessed the first sound I heard as the Meridian, but
it was the Museatex. As soon as I heard the next choice I immediately
recognized IT as being the Meridian and knew I’d made a mistake, but
the incorrect guess was ‘on record.’
These units simply sounded different. All the differences I’d
mentioned in my first post were easily heard in this test as well:
The Meridian’s clearer, sharper sound, better image focus, higher
highs, excellent separation of instruments etc. The Museatex’s
bigger, lusher sound and deeper, wider soundstaging, it’s smoother
sound, and a dead giveaway being it’s bigger, deeper bass.
Interestingly, the Museatex displayed more of the original acoustics
of the recording - reverb trails and all - than the Meridian. I’d
have guessed the reverse - that the Meridian with it’s ‘extended’
treble energy would evince the acoustics more, but this was not the
case. (I’d noted this long before in my subjective comparisons).
Again, not HUGE differences, but distinct and detectable.
I’ve always felt that I perceived the differences between these units
even when I’m listening from another room. Aha, another test! I knew
that I’d hooked the Meridian to the CD input and the Museatex to the
Auxiliary input. So I listened from another room, about twenty feet
from the opening of the listening room.
My FIL manned the selector and I shouted my guesses as to which input
he’d selected.
Results of TEST B:
(Out of 10 trials)
Incorrect Guesses: none
Correct Guesses: 10
Again, as mentioned in my last post, the Museatex sounded smoother,
fuller, with bigger bass, but the Meridian sounded cleaner, brighter
and tonally more convincing. I’ve always felt that if I’m paying big
bucks for a source whose attributes disappear unless I’m sitting
facing the speakers, then it’s not worth my money. I listen too often
from various rooms adjacent to my listening/living room.
Next, I brought out my wife’s Sony CDP-295 CD player. It’s
approximately seven years old, and in excellent condition. I wanted
to test it against the Meridian and the Mietner because it is held by
many people that an expensive high-end CD player will not improve on
the sonic performance of a well-made mid-level Sony player (which this
is). Unfortunately, the Sony’s output level was SLIGHTLY lower than
the Meridian/Museatex. Since the Sony had no separate volume control,
we had to match the outputs using the pre-amp volume and mark beside
the volume pot how much my FIL had to turn it to match outputs (very
little). I realize this is not as perfect a set-up as the
Mietner/Meridian test, but my FIL was very good at getting within a
couple of 10ths of a volt using the pot marks (on our test tone).
Anyway, I’ll give you the test results:
(I started lowering the trial numbers, so as not to wear out my FIL’s
patience, plus it was more cumbersome: my FIL had to open and close
both CD trays at once so I didn’t know which unit was receiving the
CD. There was a CD in each machine - so that I always heard the sound
of two drawers opening and two CDs being removed, but I couldn’t tell
if a switch was happening.)
Results of TEST C
Meridian vs Sony
(out of 6 trials)
Incorrect Guesses: 0
Correct guesses: 6
Again, very easy. The Sony sounded flat, more electronic, less
detailed, less spacious, less real. The drum hi-hat on the Sony
sounded more like white noise bursts, whereas on the Meridian, the
drumstick and texture of the high-hat was more audible and natural.
The Meridian simply exceeded the Sony in all those areas that we would
deem ‘higher fidelity.’ When distant string lines entered behind the
singer, on the Sony
they were ‘colorless,’ flat, thin - I would not confidently tell if
they were synthetic, sampled or real strings.
On the Meridian the strings were better separated in the mix, with
tone and texture that immediately said ‘real strings.’ Subtle, but to
me, significant.
Next we tried the ‘Listening from the other room’ test. Here’s the
results:
TEST D
Meridian/Sony
(out of 12 trials)
Incorrect Guesses: 0
Correct Guesses: 12
‘nuf said.
Next up: Sony VS Museatex.
I figured this one would be the hardest, because there wasn’t anything
like the Meridian’s extended treble to cue me - both the Sony and the
Museatex seemed to have the same rolled off treble (relative to the
Meridian). However, once I heard the difference between them (even
level-matched) I could tell each apart.
Sony vs Museatex
(out of 8 trials)
Incorrect Guesses: 1
Correct Guesses: 7
Same thing as the first test - heard the Sony first and thought "
rolled off treble, must be the Museatex." But after hearing the
Museatex I recognized it and could identify it accurately. The
Museatex sounded fuller, lusher, bigger soundstage by far (relatively)
and deeper bass. The main thing is the Museatex just sounded more
rich, more ‘real’ in a way that allowed me to listen to the performers
as if I were eavesdropping in the studio. The Sony just sounded too
one-dimensional and electronic in comparison.
Must I say again - subtle sonic differences, big subjective effect.
OK, so there you are. I’ve tried to describe the tests exactly as
they occurred, with no fudging. If I wanted to confirm my biases I’d
just invite an audio buddy over to talk about how great my components
are. Instead, I wanted to confront my subjective biases to see if
they hold up in an objective test. Gimme the truth, I can take it, I
swear.
Any comments on these shenanigans?
For me, digital is a solved problem, I use a benchmark DAC, and I don’t think about ”sonic differences between CD players/DACs.”
However, as I’ve mentioned before, I had some experience in the late 90s perceiving what seemed to be some obvious differences between some CDPs and a DAC. I’d been able to borrow various gear from friends or dealers and at one point I had the Meridian 508.20 CDP, a Museatex Bidat DAC (which had a volume control), and the good quality Sony CDP my wife and I had been using up to that point.
I was surprised at what seemed to be very distinct Sonic differences between each of them. But since it was general in my position, I shouldn’t hear a differences between competently designed digital gear, and understanding the possibility of bias effects, I did a blind test and I was able to easily identify each of them.
To get some criticism and feedback I presented the feedback and results online to a place where a number of engineers and “objectivists” hung out. Anybody here who goes back far enough might recognize names like Arny Kruger, Howard Festler, Stewart Pinkerton, and I believe our own J.J. was there as well.
I received some excellent feedback and constructive criticism as to how I could tighten up the test (which included switching from matching volume level using a RadioShack metre, to using a voltmeter to measure the speaker terminals to match output).
I redid my tests and presented those results again to the gang.
I’m just going to copy and paste my old post describing my attempts at the blind test and the results.
It might be mildly entertaining for some to read about an audiophile attempting blind testing at home, and also why it’s not exactly an easy thing to pull off. And of course I leave opinions of the test results up to the reader:
—————————-
—————————-
I wrote the original post entitled "My BLIND TESTof CD Players -
here’s the results…."
That post detailed my experience in hearing differences between my
Museatex DAC and my Meridian 508.20 in a blindtest that I
constructed. Many helpful types pointed out to me that the Radio
Shack meter I used was not the proper device to employ if I wanted to
match output levels with real accuracy. "Get thee a voltmeter and do
it right" they said. And so I went forth, and lo, I did find a
voltmeter.
I borrowed a voltmeter from the tech department at work (I work at a
large sound editing/mixing facility).
The sound technicians showed me how to work it, and best deploy it,
for my test.
I had been working up a sweat imagining having to beg my spouse to
help me in another audio-nerd test. In a stroke of luck, my
father-in-law happened to be around to help me do these tests. He’s a
classical music nut and owns a beautiful-sounding music system. He is
also a hard-nosed engineer, and does not bother listening to
components that should sound the same (re: amps, CD players etc.) -
his late 80’s CD player and amp are doing just fine, thank you.
Luckily he’s more sympathetic to the cause of good sound, and I was
able to squeeze many more test scenarios out of this session than in
the first one.
Right then. The test.
I chose a comfortable volume at which to listen, and used the
voltmeter to measure the level at the speaker terminals. My speakers
measured a few tenths-of-a-volt different in level, so I used my
pre-amp’s balance control slightly to even them out. I set both the
Meridian and the Museatex DAC to 1.72 volts. As in the previous test,
they measured the same, and I could not perceive a volume difference
between the two units.
The method we used was the same as the last test (as this was deemed
acceptable by the critics). He did not know which DAC was connected
to which inputs. I told him not to simply switch between the two
sources constantly, but to make the switching unpredictable and
‘random.’ A sans music pre-test showed that, with a few ‘fake’ switch
movements of the source selector in between source selections, I could
not reliably guess which source had been selected. I was blind-folded
and we did not talk to each other during the test, except when I said
"switch." My FIL (father-in-law), kept track of my correct and
incorrect answers.
Here’s the score for TEST A:
Meridian vs Museatex
Out of 18 trials
Incorrect Guesses: 1
Correct Guesses: 17
For some reason I guessed the first sound I heard as the Meridian, but
it was the Museatex. As soon as I heard the next choice I immediately
recognized IT as being the Meridian and knew I’d made a mistake, but
the incorrect guess was ‘on record.’
These units simply sounded different. All the differences I’d
mentioned in my first post were easily heard in this test as well:
The Meridian’s clearer, sharper sound, better image focus, higher
highs, excellent separation of instruments etc. The Museatex’s
bigger, lusher sound and deeper, wider soundstaging, it’s smoother
sound, and a dead giveaway being it’s bigger, deeper bass.
Interestingly, the Museatex displayed more of the original acoustics
of the recording - reverb trails and all - than the Meridian. I’d
have guessed the reverse - that the Meridian with it’s ‘extended’
treble energy would evince the acoustics more, but this was not the
case. (I’d noted this long before in my subjective comparisons).
Again, not HUGE differences, but distinct and detectable.
I’ve always felt that I perceived the differences between these units
even when I’m listening from another room. Aha, another test! I knew
that I’d hooked the Meridian to the CD input and the Museatex to the
Auxiliary input. So I listened from another room, about twenty feet
from the opening of the listening room.
My FIL manned the selector and I shouted my guesses as to which input
he’d selected.
Results of TEST B:
(Out of 10 trials)
Incorrect Guesses: none
Correct Guesses: 10
Again, as mentioned in my last post, the Museatex sounded smoother,
fuller, with bigger bass, but the Meridian sounded cleaner, brighter
and tonally more convincing. I’ve always felt that if I’m paying big
bucks for a source whose attributes disappear unless I’m sitting
facing the speakers, then it’s not worth my money. I listen too often
from various rooms adjacent to my listening/living room.
Next, I brought out my wife’s Sony CDP-295 CD player. It’s
approximately seven years old, and in excellent condition. I wanted
to test it against the Meridian and the Mietner because it is held by
many people that an expensive high-end CD player will not improve on
the sonic performance of a well-made mid-level Sony player (which this
is). Unfortunately, the Sony’s output level was SLIGHTLY lower than
the Meridian/Museatex. Since the Sony had no separate volume control,
we had to match the outputs using the pre-amp volume and mark beside
the volume pot how much my FIL had to turn it to match outputs (very
little). I realize this is not as perfect a set-up as the
Mietner/Meridian test, but my FIL was very good at getting within a
couple of 10ths of a volt using the pot marks (on our test tone).
Anyway, I’ll give you the test results:
(I started lowering the trial numbers, so as not to wear out my FIL’s
patience, plus it was more cumbersome: my FIL had to open and close
both CD trays at once so I didn’t know which unit was receiving the
CD. There was a CD in each machine - so that I always heard the sound
of two drawers opening and two CDs being removed, but I couldn’t tell
if a switch was happening.)
Results of TEST C
Meridian vs Sony
(out of 6 trials)
Incorrect Guesses: 0
Correct guesses: 6
Again, very easy. The Sony sounded flat, more electronic, less
detailed, less spacious, less real. The drum hi-hat on the Sony
sounded more like white noise bursts, whereas on the Meridian, the
drumstick and texture of the high-hat was more audible and natural.
The Meridian simply exceeded the Sony in all those areas that we would
deem ‘higher fidelity.’ When distant string lines entered behind the
singer, on the Sony
they were ‘colorless,’ flat, thin - I would not confidently tell if
they were synthetic, sampled or real strings.
On the Meridian the strings were better separated in the mix, with
tone and texture that immediately said ‘real strings.’ Subtle, but to
me, significant.
Next we tried the ‘Listening from the other room’ test. Here’s the
results:
TEST D
Meridian/Sony
(out of 12 trials)
Incorrect Guesses: 0
Correct Guesses: 12
‘nuf said.
Next up: Sony VS Museatex.
I figured this one would be the hardest, because there wasn’t anything
like the Meridian’s extended treble to cue me - both the Sony and the
Museatex seemed to have the same rolled off treble (relative to the
Meridian). However, once I heard the difference between them (even
level-matched) I could tell each apart.
Sony vs Museatex
(out of 8 trials)
Incorrect Guesses: 1
Correct Guesses: 7
Same thing as the first test - heard the Sony first and thought "
rolled off treble, must be the Museatex." But after hearing the
Museatex I recognized it and could identify it accurately. The
Museatex sounded fuller, lusher, bigger soundstage by far (relatively)
and deeper bass. The main thing is the Museatex just sounded more
rich, more ‘real’ in a way that allowed me to listen to the performers
as if I were eavesdropping in the studio. The Sony just sounded too
one-dimensional and electronic in comparison.
Must I say again - subtle sonic differences, big subjective effect.
OK, so there you are. I’ve tried to describe the tests exactly as
they occurred, with no fudging. If I wanted to confirm my biases I’d
just invite an audio buddy over to talk about how great my components
are. Instead, I wanted to confront my subjective biases to see if
they hold up in an objective test. Gimme the truth, I can take it, I
swear.
Any comments on these shenanigans?
Last edited:
An AD process begins with taking samples, “snap shots” at sample time, and converting the value to a digital number. It all has to be done before the next sample time. Accurate conversion requires that the “snap shots” samples will be accurate. Sample time error is another issue.
I started as an analog circuit designer. That is my foundation. I am a professional electron pusher and manipulator. The signal flows through the circuits. It takes time from input through all the devices and circuits to the output. So I have to track the signal “travel” through the circuits at many points. I have to think in the time domain, and all of it for making each sample accurate.
I don’t view music (air or voltage variations over time) as simple sine waves with different amplitudes and phases. I view the music waves as containing a lot of energy of a fundamental and harmonics, but also other energy. Music is not a steady tone. The “envelop” (such as attack of a drum, decaying piano note…) contains energy. The math becomes more complicated. It is no longer Fourier series and simple FFT’s, it is the Fourier integral. I use it for circuit design and simulation, but it is not needed for testing because the circuit can be verified in the frequency domain (with sine and multiple sine based tests). I too relate to music in the frequency domain. But I design to make each sample accurate (for audio bandwidth and dynamics) in order to get the results in the frequency domain. It is all connected.
If you want accuracy (I will expand more later), you need your conversion of each sample to be accurate. It works for MRI, for digital scope, postal weighing scale, for audio…. I am glad they test the MRI AD’s carefully…
I started as an analog circuit designer. That is my foundation. I am a professional electron pusher and manipulator. The signal flows through the circuits. It takes time from input through all the devices and circuits to the output. So I have to track the signal “travel” through the circuits at many points. I have to think in the time domain, and all of it for making each sample accurate.
I don’t view music (air or voltage variations over time) as simple sine waves with different amplitudes and phases. I view the music waves as containing a lot of energy of a fundamental and harmonics, but also other energy. Music is not a steady tone. The “envelop” (such as attack of a drum, decaying piano note…) contains energy. The math becomes more complicated. It is no longer Fourier series and simple FFT’s, it is the Fourier integral. I use it for circuit design and simulation, but it is not needed for testing because the circuit can be verified in the frequency domain (with sine and multiple sine based tests). I too relate to music in the frequency domain. But I design to make each sample accurate (for audio bandwidth and dynamics) in order to get the results in the frequency domain. It is all connected.
If you want accuracy (I will expand more later), you need your conversion of each sample to be accurate. It works for MRI, for digital scope, postal weighing scale, for audio…. I am glad they test the MRI AD’s carefully…
I may have over stressed the time domain. Of course I am very aware of how circuits and components relate to frequency and to audio. I am also aware of what needs to be accurate from psycho-acoustic stand point. I am just trying to point out that there is another angle to viewing conversion. Time is such an important aspect for design. The sample and convert model is general description of the theory. In reality we Sigma delta and or oversampling should yields the same outcome, accurate samples. Most of the industry relates to conversion via measurements, most often in the time domain. If you feed an AD a ramp, it turns to a stair case. We want the steps to be equal (differential linearity) and the ramp slope to be constant (integral linearity). It is not easy to measure. It is easier to make pure sine waves and to use filtering to measure.
I am talking about design. Let me know if it is boring or if I am getting of topic to much. The forum seems to have people with different experience and interests.
I’m not sure I understand this comment, because the “corner cases” I’ve heard are usually something like patterns of sustained or intermittent power delivery across odd bandwidths in an amplifier.
My general point is that in this business (audiophilia/ “high end”) people take the germ or abstraction of a true physical property or discovery and extrapolate it into an elaborate rationalization for abandoning objective measurements entirely. Another one that comes to mind is the way high end cable vendors will borrow quantum anomalies from long-distance/high power transmission as if they are remotely applicable to the six feet of copper from their amp to their speakers. Hence the “Motte-and-Bailey” cartoon.
Last edited:
The references to 'high end' you're seeing here, are snark. They refer to the consumer 'high end ', i.e., nonsense pushed historically by The Absolute Sound and Stereophile, 'high-end' boutique retailers, and now, in the Internet era, also by lots of online slop sites and video content makers. ASR stands athwart all that noise, yelling NOPE.
It has nothing to do with the stellar work you've done, which is truly 'high end'.
Though of course there are some 'pros' in audio production suites who swear by audio myths and nostrums. My take on them is that their insistance that, e.g., everything must be done at 192/24, or that Shakti Stones must be used, has no practical effect whatsoever on the audio I hear, negative or positive, so, let them indulge their silly belief, as long as the recording I buy sounds good.
Last edited:
The asserted differences, perceived under conditions that aren't conducive to accuracy.
Was the comparison level-matched? Was it double-blind?
Measurements can also be dispositive. They can tell if one or both DACs are broken...or really badly designed.
Welcome to ASR. Though you've been here for a couple of years already? Do you not know the hurdles that need to be jumped for your claim to be taken as a 'known', here?
As an engineer by trade, you should know that making a claim like yours on a science-focused forum means you have to come 'armed for bear'. You need to show the 'specs'.
IOW, you need to state upfront exactly how you compared your DACs, detailing your 'methods', not dribbling them out in response to questions.
Behaving otherwise signals either ignorance, a lack of seriousness, or at worst, bad faith (aka trolling).
Stop right there.
It's not an 'on the other hand' thing. No one here is claiming *electromechanical transducers* that 'measure quite similar' can't be audibly distinguished under controlled conditions. How similar and under what testing regime matters a lot there. So why even bring it up?
Unfortunately, the rest of your post is the usual bargle -- anecdota + reference to questionable online tests -- that all needs far more interrogation than it's worth, though I see ASR has tried.
Video had so much further to go.
I honestly don't know what' you're talking about.
It's child's play to implement a 2.1 system that 'anyone can handle' and is 'affordable'.
I assume 2.1 means, two 'main' speakers, and one subwoofer.
What use case are you thinking of? Something portable?
This is a common idea in some humanities or 'sociology' departments, but it's like saying engineering that aims to make bridges that work, rather than fall down, is an ideology
I was attracted to this forum when I saw the thread about measurements, so important to my professional life. But I am not sure that I can contribute much to the conversations about hearing and DBT comments. I have been listening to my customer’s opinions for many years. If three independent people tell you the same thing, it is wise to listen, (feedback can be about anything, the sound, LED brightness, missing feature and more).
But I think it is difficult for “ear people” (even well trained ears) to correlate the “outcome” to the cause. For example a high pitch issue turned out to be a lower frequency response issue. It is all supported by math, but the “ear person” hears high pitch, reports about high pitch issue… So once again, music is more complicated than “sine waves with harmonic and phase”. It is impractical to view (or test) an orchestral as individual fundamentals and harmonics. It is more practical for me to view music as “anything withing parameters related to human hearing” (or 96KHz sampling for your dog’s benefit). Music can be “Anything” leads to being able to accommodate a full-scale random signal with flat frequency response (or finger nails scratching a black board). But testing and understanding product behavior can be done with sine waves (and multiples), jitter measurements and more.
I fully support the need for extensive measurement, but I don’t think I am a good fit here. I looked at other threads, talking about so many different products. I am a converter guy, and I don’t look at other manufacturer products… I like to talk about product design, and tell stories, probably of little interest to most people. That is cool. You do have some real expertise here. JJ’s is a “rare bird”, a research scientist, deep thinker with very strong high-level math.
But I think it is difficult for “ear people” (even well trained ears) to correlate the “outcome” to the cause. For example a high pitch issue turned out to be a lower frequency response issue. It is all supported by math, but the “ear person” hears high pitch, reports about high pitch issue… So once again, music is more complicated than “sine waves with harmonic and phase”. It is impractical to view (or test) an orchestral as individual fundamentals and harmonics. It is more practical for me to view music as “anything withing parameters related to human hearing” (or 96KHz sampling for your dog’s benefit). Music can be “Anything” leads to being able to accommodate a full-scale random signal with flat frequency response (or finger nails scratching a black board). But testing and understanding product behavior can be done with sine waves (and multiples), jitter measurements and more.
I fully support the need for extensive measurement, but I don’t think I am a good fit here. I looked at other threads, talking about so many different products. I am a converter guy, and I don’t look at other manufacturer products… I like to talk about product design, and tell stories, probably of little interest to most people. That is cool. You do have some real expertise here. JJ’s is a “rare bird”, a research scientist, deep thinker with very strong high-level math.
That's it, in a nutshell.
People are prone to attribute perceived audible 'difference' to whatever they 'think' causes it...and the easiest thing to pin it on, is the fact that DAC A is not DAC B.
I was present in very few listening tests (it takes a lot of travel). But the DBT idea is based on comparing units with ALL other variables the be the same in all tests. Some of the mastering people listen to music all they long. Unlike normal listeners, they seem to be “trained” to focus on one aspect or another. I have to accept their expertise. Yes, people can be wrong and often are. So I look for some consensus, preferably three or more independent sources in agreement is pretty solid. (I first go to a couple of ears that seem to get it right most often).
So I can’t add much. I too am listen to opinions (my hearing aids are not up to the task). But I know how to approach product design, as an old man with 8KHz bandwidth. Here is a story:
I was very old, facing the fact that it may be my last design. My DA does 192KHz but the AD did not. Also, the high AD latency restricted it's use for mostly mastering, so I needed a redesign. My converters are for professional use with appropriate features. My customers are top notch studio and facilities. So how good should the converter be? What is the appropriate resolution? The answer was simple. I am going to do the best I can. There was (hopefully) no time limit, no cost constrained, no boss breathing down my knack. I concentrated on accuracy beyond hearing capabilities.
Some pro applications (digital archives) have clear benefits (with no room for rational arguments) from using a very high resolution AD, but that is another topic. I designed the AD for normal mastering, recording and mixing. I was surprised to find out so much archiving (conversion to digital) activity.
Roland68
Major Contributor
I'm talking about something very simple.
A 2.1 system where the speakers are relieved of the low frequencies, the subwoofer of the high frequencies, separately adjustable crossover frequencies, adjustable phase, crossover slope, volume, ideally with contour adjustment, etc.
All of this integrated into DACs or desktop-sized DAC/amplifier combinations, as has been the case with AVRs for over 30 years.
The WiiM would be one of the standard solutions, but not everyone wants or needs a WiiM, and often it doesn't fit into the overall concept/system. Of course, there are MiniDSPs and a few more expensive amplifiers that have this integrated.
A recent project on ASR demonstrated that all this functionality can be implemented in a Raspberry Pi Pico for just a few euros/dollars.
If you know of anything, please post it. The numerous 2.1 advice threads constantly show how few usable solutions there are.
Everything not bolded has been available in mass market gear for decades. 'Contour adjustment', if I understand you correctly, is more recently available via some room eq DSPs, like the Audyssey app. Is there a particular reason you need control of crossover slopes? A 2.1 desktop system isn't rocket science.
I guess it's a matter of how big the desktop is. Do you need the head unit to be literally on the desktop? An AVR is out of the question?
I also have distinct thoughts about this, but this should move to its own DIY thread or something, maybe?
Especially with the advent of low-end pi's and switching amps. Some things strike me as odd here, why have variable crossover rates? Why not brutally fast FIR's? They sound fine, and protect drivers like nothing else in the world.
A case against High bit rate digital audio
AD conversion requires an analog anti-aliasing filter to provide a sufficient attenuation and a sharp transition (from say passing 20KHz audio to full rejection at Nyquist). Higher sample rates (higher Nyquist) provide for practical realizable filter (gentler slope).
DA conversion requires an analog anti imaging filter (to remove the extra high frequency in the reconstructed signal). Similar to the AD converter, higher sample rate is critical for achieving good filter. Also, higher sample rate corrects for high frequencies amplitude loss following sin(x)/x amplitude vs. frequency response.
The hard way or the easy way to do it? Let's compare 384Khz and 96KHz sample rates.
A 384KHz converter offers a theoretical 192KHz bandwidth, say 160KHz of real performance. It is about X8 the human hearing range (I assume here 20-20KHz). The microphones and speakers are not (and should not) operating over such frequency range. The 384KHz conversion solves the analog filter problem, but the extra bandwidth that comes with it is not needed.
The 384KHz converter generates X4 more data than 96KHz. It slows down upload, download and internet. The file size requires X4 memory. Also, the hardware to connect digital signals between units (transmitter, cable and receiver) is almost 50MHz. (AES 128-bit encryption). But it solves the filter issue.
Converting at 96KHz digital rate is around 12.28MHz (AES 128-bit encryption). There is more time (X4) to convert each sample, offering more accuracy. But it has only 1/4 of the samples. It needs more samples to make the DA anti imaging be the same as the 384KH case.
That is what up sampling does. It computes the “missing” samples inside the DA from the 96KHz data stream. The digital computation is very precise. The up sampling makes the filter THE SAME in both cases.
Both 384KHz and the up sampled 96K X 4 = 384KHz have the same filter requirement.
The 384KHz case offers a lot of unused bandwidth while complicating the hardware, bigger files, more memory, slower internet…
The 96KHz covers more than 40KHz (twice the hifi range) with %25 file size and a simpler (slower) digital audio interface. It does not extend the bandwidth that you don’t need!
Again, microphones and speakers are not design to accommodate 160KHz (X8 hifi range of 20-20KHz). The 96KHz is more than enough.
The above is a specific example. Try to compare 48KHz to 768KHz. It is like buying 16 cars to never use 15 of them. And the operation of that one car is complicated (768K*128 = 96MHz digital audio hardware).
The concepts of up sampling and oversampling are well known and utilized by converters for dozens of years. There is no advantage in creating, storing or transmitting all that data when the DA can calculate the missing sample value from the lower sample rate, and do so precisely. Why 384KHz and 768KHz for audio? Beats me. I think that 96KHz is more than enough.
AD conversion requires an analog anti-aliasing filter to provide a sufficient attenuation and a sharp transition (from say passing 20KHz audio to full rejection at Nyquist). Higher sample rates (higher Nyquist) provide for practical realizable filter (gentler slope).
DA conversion requires an analog anti imaging filter (to remove the extra high frequency in the reconstructed signal). Similar to the AD converter, higher sample rate is critical for achieving good filter. Also, higher sample rate corrects for high frequencies amplitude loss following sin(x)/x amplitude vs. frequency response.
The hard way or the easy way to do it? Let's compare 384Khz and 96KHz sample rates.
A 384KHz converter offers a theoretical 192KHz bandwidth, say 160KHz of real performance. It is about X8 the human hearing range (I assume here 20-20KHz). The microphones and speakers are not (and should not) operating over such frequency range. The 384KHz conversion solves the analog filter problem, but the extra bandwidth that comes with it is not needed.
The 384KHz converter generates X4 more data than 96KHz. It slows down upload, download and internet. The file size requires X4 memory. Also, the hardware to connect digital signals between units (transmitter, cable and receiver) is almost 50MHz. (AES 128-bit encryption). But it solves the filter issue.
Converting at 96KHz digital rate is around 12.28MHz (AES 128-bit encryption). There is more time (X4) to convert each sample, offering more accuracy. But it has only 1/4 of the samples. It needs more samples to make the DA anti imaging be the same as the 384KH case.
That is what up sampling does. It computes the “missing” samples inside the DA from the 96KHz data stream. The digital computation is very precise. The up sampling makes the filter THE SAME in both cases.
Both 384KHz and the up sampled 96K X 4 = 384KHz have the same filter requirement.
The 384KHz case offers a lot of unused bandwidth while complicating the hardware, bigger files, more memory, slower internet…
The 96KHz covers more than 40KHz (twice the hifi range) with %25 file size and a simpler (slower) digital audio interface. It does not extend the bandwidth that you don’t need!
Again, microphones and speakers are not design to accommodate 160KHz (X8 hifi range of 20-20KHz). The 96KHz is more than enough.
The above is a specific example. Try to compare 48KHz to 768KHz. It is like buying 16 cars to never use 15 of them. And the operation of that one car is complicated (768K*128 = 96MHz digital audio hardware).
The concepts of up sampling and oversampling are well known and utilized by converters for dozens of years. There is no advantage in creating, storing or transmitting all that data when the DA can calculate the missing sample value from the lower sample rate, and do so precisely. Why 384KHz and 768KHz for audio? Beats me. I think that 96KHz is more than enough.
