Is there more to Audio Measurements?

Status
Not open for further replies.
As I said gathering together such measurements of music-like test signals, like multitone may well show correlation between what's seen on the measurements & subjective listening.

It would be interesting if the collection of multitone measurements on that site showed this correlation but I don't have experience with the devices tested to make this correlation. For instance, I don't know if all the DACs that show IMD products rising towards HF subjectively sound sibilant? FiiO X1, FiiO X3 II, FiiO X5 II, HiFiMan HM 700, HiFiMan HM 802 HD, etc

Pinknoise listening test might be an interesting way to access this?
 
It seemed possible at the time but the multitone testing did not separate anything audible for me. As I already mentioned it does not exercise the noise modulation mechanisms that Martin was talking about.

Since you are familiar with Dave Griesinger I assume you might have read his paper on speaker distortion (and a few other things). This is where I began to question the actual content of supposed 24/96 files people were putting up for listening tests. The last slide.

Conclusions: A/B tests
No evidence was uncovered in this study that would invalidate rapid, blind, A/B tests as the gold standard for audio research.

But the possibility remains…
Particularly in the study of room acoustics
intelligibility, muddiness, and envelopment all may depend on the time period devoted to listening to a particular acoustic signal
 
When 24/96 sounds better it seems doubtful it is simply because of content. Rather, some D/A converters may sound better at higher sample rates. Could have something to do with how they respond to jitter as a function sample rate or something like that. The best data converters don't sound much, if any different with sample rate changes, IME. That being said, most are far from the best, so for most practical purposes higher sample rates may actually sound better.

Also, as some may be aware, jitter is one of those things that can produce very non-stationary distortion and noise. Probably will require specialized tests to accurately quantify, again IMHO.
 
It seemed possible at the time but the multitone testing did not separate anything audible for me. As I already mentioned it does not exercise the noise modulation mechanisms that Martin was talking about.
I never said that multitone testing would reveal all things audible, just that it might be one measurement that better correlates to subjective listening.

As regards ESS DACs, John Siau's answer to this question here might shed some light or maybe not?
Q: A lot of companies who are using ESS dacs have a dramatic rise in smpte imd at around -25dbfs, however the benchmark dac 3 does not have this issue. Can you share any insight into what those companies are doing wrong?
A: This IMD (and THD) is a direct result of omitting a differential amplifier after the output of the chip.......
All D/A converter chips produce significant common-mode distortion on the balanced outputs. This unwanted common-mode distortion is easily removed with a differential amplifier. If it is not removed, the result is IMD.

Since you are familiar with Dave Griesinger I assume you might have read his paper on speaker distortion (and a few other things). This is where I began to question the actual content of supposed 24/96 files people were putting up for listening tests. The last slide.
Yes, Gresinger's conclusion about Ashira's paper seemed reasonable - not sure what your point is, though?

If it's in relation to A/B testing Vs long term testing, his text seems appropriate "intelligibility, muddiness, and envelopment all may depend on the time period devoted to listening to a particular acoustic signal" but I wouldn't restrict what he says to just evaluating room acoustics, I believe long term listening may reveal aspects of a systems performance which are not room related but inherent aspects of the playback system.
 
If you are interested in DAC's you might want to check out Martin Mallinson's (from ESS and an old friend) youtube presentations on artifacts in DAC's.
Thanks Scott. I've watched one and am half way thru another. He's amazingly good at explaining this stuff, his knowledge is "historic" to say the least. 🙂

I designed an ESS Sabre DAC that almost made it to market. I liked it, very smooth and gentle sound, no rough edges. The prototype at RMAF didn't do well, seemed like it wasn't aggressive enough for most people. Might have been my implementation of it. I had better luck with Cirrus Logic chips.
 
But Mr. Marsh and others claim Quad speakers were (and still are) particularly low distortion speakers. Such confusion.

They even have measurements to prove it.

Generally, Joe is correct....IMO.... but there are a few ways to reduce any speakers' distortion..... listening in the near field will limit the cone excursion a lot for a given spl and all but eliminate the room from the sound. The old QUAD speaker and amp? I don't know... but the "new" QUAD speakers are very low in thd (0.1%). BTW several planar and push-pull ESL headphones may be well below this thd level.

One surely needs low distortion from speakers/headphones to judge any other gear.


THx-RNMarsh
 
Last edited:
Generally, Joe is correct....IMO.... but there are a few ways to reduce any speakers' distortion..... listening in the near field will limit the cone excursion a lot for a given spl and all but eliminate the room from the sound. The old QUAD speaker and amp? I don't know... but the "new" QUAD speakers are very low in thd (0.1%). BTW several planar and push-pull ESL headphones may be well below this thd level.

One surely needs low distortion from speakers/headphones to judge any other gear.


THx-RNMarsh

I think it's a common fallacy that speaker distortion will swamp most other distortions - that's not my experience - electronics tend to be more egregious & if you read that Greisinger presentation, he also concludes

"The various tweeters tested – 3 metal dome tweeters and one soft dome tweeter – produce insignificant amounts of intermodulation products below 20kHz when driven by ultrasonic signals.
Amplifier distortion can produce distortion products below 20kHz that are audible (with difficulty) in the absence of other signals below 20kHz."​
 
Last edited:
"The various tweeters tested – 3 metal dome tweeters and one soft dome tweeter – produce insignificant amounts of intermodulation products below 20kHz when driven by ultrasonic signals.
Amplifier distortion can produce distortion products below 20kHz that are audible (with difficulty) in the absence of other signals below 20kHz."​

A few points here, amplifiers are broad band devices, tweeters alone are not capable of generating substantial SPL at lower frequencies.

What is the meaning of amplifier distortion in the absence of a speaker to reproduce it? Also note the "with difficulty" even in the absence of any masking signal. Play 15kHz 1/3 octave noise on an LP and listen, the "garbage" is obvious.
 
A few points here, amplifiers are broad band devices, tweeters alone are not capable of generating substantial SPL at lower frequencies.
And your point is?

What is the meaning of amplifier distortion in the absence of a speaker to reproduce it?
Huh? Amp distortion can be measured directly from its outputs
Also note the "with difficulty" even in the absence of any masking signal.
So?
Play 15kHz 1/3 octave noise on an LP and listen, the "garbage" is obvious.
Are you saying that speaker distortion renders upstream distortions practically undetectable?
 
Are you saying that speaker distortion renders upstream distortions practically undetectable?

Let's make it simple, a tweeter trying to play 20kHz and 20.2kHz can't generate much 200Hz SPL by simple geometry and physics of the dome size. An amplifier has no problem.

The LP point is that a phono cartridge is also a broad band device and gladly generates lots of lower frequency IM distortions. This does not bother a lot of folks including me.

You have the wrong person if you want to discuss analytical listening, I got tired of listening to the first minute of a song 30 times over while someone fiddled with whatever a long time ago.
 
Let's make it simple, a tweeter trying to play 20kHz and 20.2kHz can't generate much 200Hz SPL by simple geometry and physics of the dome size. An amplifier has no problem.

The LP point is that a phono cartridge is also a broad band device and gladly generates lots of lower frequency IM distortions. This does not bother a lot of folks including me.
OK, maybe my quoted text was not accurate enough in the penultimate slide Greisinger states this about speakers which covered the full bandwidth, not just tweeters
"The loudspeakers tested have intermodulation distortion lower than the threshold of detection for complex tones."

You have the wrong person if you want to discuss analytical listening, I got tired of listening to the first minute of a song 30 times over while someone fiddled with whatever a long time ago.

Sure, that type of comparative listening is a bore - you must also be bored by forced choice blind listening too!
 
IM distortions of cartridges are obviously pretty similar to IM distortions of eardrums. I can imagine that some people with clogged ears or some loss of hearing would prefer a simulation of louder music by cartridges, triode outputs, loudspeakers, but I am not speaking about them, I mean we should forgive some distortions that our hearing can't recognize as such, in order to be happy, and optimize our systems such a way so nasty sounding in dynamics distortions would be minimized.
 
I think it's a common fallacy that speaker distortion will swamp most other distortions - that's not my experience - electronics tend to be more egregious [/INDENT]

I would agree with that also. To a point and then lower distortion is needed again.

It is the just detectable change or delta which can be heard.

The well known two closely spaced tones...one lower in level than the other --- both individually can be heard until one dominates and the other cant be heard as a distinctly different/separate tone. It seems missing. (Masking). But do the same thing and ask a different question... do the 2 tones (one masked) sound exactly the same as the single louder tone, alone? No.


THx-RNMarsh
 
Last edited:
It's best not to extrapolate the results from simple signals - both for measurements & listening.

People try to do this all the time & we can see in measurements that using test signals which better represent the complex signal structure of music, show more about the non-linear distortions of what is often considered blameless electronics. Even multitone & pinknoise test signals are still far away from the final complex signal (music) that the playback system is expected to handle.

The same applies to listening but even more so - auditory perception is not so concerned with frequency, amplitude - it's job is to construct the auditory scene which makes the most sense from all the concurrent waveforms that are hitting the eardrum moment to moment. In other words to associate different frequencies amplitudes timing of various signals into auditory objects which track through time. We are born with a certain innate faculty which is designed to accomplish this but maybe 90% of it is learned from exposure to real world sounds & verified by our other senses, mainly sight (the same applies to the visual perception, btw). So this 90% that determines our auditory perception is about the ongoing relationship & patterns within the stream of audio waves arriving at our eardrums. We are much more sensitive to a subtle break in a pattern then we are to other changes so auditory testing with tones is a very basic form of listening tests & I believe, really has mislead us in our determination of what's audible/noticeable & what's not.

A lot of what's noticeable are not direct effects but secondary/tertiary effects on the sound. So instead of looking for exact & identifiable spot differences, we should be looking at a far broader canvas - the patterns & interrelationship within the complex waveform.

As RNMarsh says "ask a different question... do the 2 tones (one masked) sound exactly the same as the single louder tone, alone? No. "
 
Status
Not open for further replies.