Does DSP belong in State of the Art Systems?

[Bjorn]

How does the mass enter into things?
At a higher frequency the driver is not moving much and the force on the air seems to be the main thing stopping the cone from moving?

Or are we talking subwoofers and woofers?

Even with those I regularly see that they play 20Hz to 1kHz. The SPL should drop with frequency, as the mass increases.
But to hit the higher frequency implies the motor is putting out force.

Lower mass=equals lower distortion for mids and highs. This is well known and measurable. It's a big topic where I would encourage you to look at some studies/paper if you want to further dive into the topic. The key is lack and of resonances and the relation to movement and also the cabinet. If you listen to a large wide band driver (for instance 3" to 5") in the highs and compare it to a small tweeter it's very evident despite that the force and movement in the highs are small. But directivity also comes into play.

Distortion is also a matter in the bass and I gave an example of horn loading a woofer, which decreases the distortion compared to front firing. It took four front firing woofers to get close to the pereceived low modulation distortion of two horn loaded 12" woofers in a comparison Klipsch did.

 

[Bjorn]

Transparency in electronics is quite straight forward. It's measurements vs listenings tests of threshold of distortion. Threshold of distortion will vary of course depending on areas like hearing, age, trained vs non trained, quality of speaker/headphones, etc. If signal out is close enough to signal in, in relation to distortion, the component can be considered transparent.

Transparency is speakers and acoustics on the other hand isn't straigh forward at all. In every speaker design there will be compromises that needs to be made. And a room will also add some signature, no matter what treatment is used.

 

[microstrip]

Transparency in electronics is quite straight forward. It's measurements vs listenings tests of threshold of distortion. Threshold of distortion will vary of course depending on areas like hearing, age, trained vs non trained, quality of speaker/headphones, etc. If signal out is close enough to signal in, in relation to distortion, the component can be considered transparent.

Yes, apparently it looks quite straight forward. But the point is that we still do not have quantitative data that can applied to our current high end stereo systems. We can not apply the data coming from old studies carried with mono tones to our debates. Since long the research community consider stereo a non interesting subject and actual stereo developments were carried by the industry, that has little interest in formal research and publicizing real science.
Research on digital was the only subject that had some activity after the 80's.

Surely if someone considers that all decently designed amplifiers and cables sound the same things are simple and straight forward - but IMHO this is not the opinion of the large majority of WBF members.

Transparency is speakers and acoustics on the other hand isn't straigh forward at all. In every speaker design there will be compromises that needs to be made. And a room will also add some signature, no matter what treatment is used.

Yes, the speaker/room system can't be transparent in terms of stereo - the system relies on speaker sound reflections that can't be standardized. At best we use the word in a subjective way, dependent on listener experience and preference.

 

[Bjorn]

Surely if someone considers that all decently designed amplifiers and cables sound the same things are simple and straight forward - but IMHO this is not the opinion of the large majority of WBF members.

Most amplifiers don't have low enough measurable distortion to be inaudible according to our present knowledge about the threshold. So in other words, most amplifiers aren't transparent either.

 

[microstrip]

In your position I can see that this might be mysterious. As any competent designer knows there's nothing mysterious about it at all. If I were to offer you advice (and this based entirely on this brief exchange so could be way off base) I would advise to dump the made-up story that perception is all that difficult or mysterious. None of it is.

Humans all have very similar perceptions and in this regard true 'Golden Ears' are really rare. If we didn't use the same hearing perceptual rules, the science and art of audio would not exist. There is a reason deciBels exist, why bandwidth has to be at least 20-20KHz and so on. I'm not going to go into all the perceptual rules here but there are a lot of them and we all have them in common unless our hearing is damaged/defective.

Yes, but also lot of differences between the individual hearing systems. Rules are created statistically for the large majority - it is why these studies must include a large population.
And unfortunately people often refer to "perceptual rules" but once we ask for them and links to studies documenting them the subject abruptly ends without this "lot" of rules.

So this could become a conversation going down some existential rabbit hole which is entirely unnecessary. Your comment about 'I doubt that has been demonstrated.' is false- its been demonstrated many times; everyone on this thread including yourself has been witness to such, unless you have a difficulty distinguishing between that which is being perceived and the perceiver. Or- I grossly misunderstand what you are trying to say.

You might ask how I know this. My response is go read the reviews of our equipment. One thing that is common to all of them is that our equipment has that character of being 'transparent'. I know what transparency is because its my job to know how to design equipment that is just that and to know when I hear it.

Since long OTLs have created the aura of being "transparent" because they do not have an output transformer. It is natural that reviewers focus on this word when reviewing an OTL - this happens in 99% of the OTL reviews I have read. I have owned several OTLs - Futtterman, Technics 20A, Graaf and Atmasphere M50 and M2. I can assure you they sounded quite different from each other, but all reviewers raved about their transparency.

Just to point that IMO reviewers and manufacturers texts can't be a reference on debates about transparency. They are just opinions.

(Whether audiophiles like it or not there is science to audio and good measurements are paramount to good performance/good sound. But you have to know what is important! Daniel Recklinghausen said it very well and his words are as true to day as they were ages ago!)

I can't see what words such as "If it measures good and sounds bad, it is bad. If it Measures bad and sounds good, you've measured the wrong thing." add to this debate. But I would love to read good science that can applied to measurements of current SOTA electronics - most I know about only applied to electronics that had much higher distortions and noise than current SOTA.

I've advised many time to make recordings yourself so you were present at the musical event and then produce media like LPs or CDs so you can play those recordings on various systems. You'll know right away what you're listening to- how well it plays bass, can you make out the smaller details, that sort of thing.

I have been present to a few performances that I have excellent recordings. It is interesting, but it also carries a risk - when listening to reproduction we focus our attention on the details that we have perceived at the real performance and ignore many others. IMO not a good test for transparency (whichever it means for each of us) . BTW, I find many people just refer to transparency as a synonymous of detailed.

When the recording equipment fools you into thinking that something you just heard was real and not coming through the headphones or speakers then you know you are on the right track. Microphones, mic preamps and headphones have been there for a really long time. Most of the transparency is lost when the signal is committed to media but even then you can still make a system sound so real that it can on occasion fool you so completely that you are convinced someone broke into your home and is singing/playing along with your stereo before they axe you to death. Its as spooky as it sounds.

Any system can fool us occasionally on some specific recording. The point is creating a system that fools us most of the time with the music we listen. And our definition of "fooling" and what is needed to reach it can differ more than our hearing!

 

[microstrip]

Most amplifiers don't have low enough measurable distortion to be inaudible according to our present knowledge about the threshold. So in other words, most amplifiers aren't transparent either.

Can we know what you consider "our (?) present knowledge about the threshold"? And which little number of amplifiers can be bellow this threshold?

 

[Bjorn]

Can we know what you consider "our (?) present knowledge about the threshold"? And which little number of amplifiers can be bellow this threshold?

Something between -110 and -120 dB for FFT, IMD, THD+N vs frequency, jitter and crosstalk would be sufficiently low. And frequency response within 0.1 dB. Researchers would have less strict numbers, but I think it's good to be on the certain side.

Obviously one needs to consider the needed power for the speakers, listening distance and SPL as well.

 

[microstrip]

Something between -110 and -120 dB for FFT, IMD, THD+N vs frequency, jitter and crosstalk would be sufficiently low. And frequency response within 0.1 dB. Researchers would have less strict numbers, but I think it's good to be on the certain side.

Obviously one needs to consider the needed power for the speakers, listening distance and SPL as well.

Thanks. What is the source of your IMHO too general numbers? Just your own opinion?
Do you also apply this criteria to digital sources?

 

[Bjorn]

Thanks. What is the source of your IMHO too general numbers? Just your own opinion?
Do you also apply this criteria to digital sources?

Based on the studies and articles I have read over the years and I've lowered certain numbers some to be certain. Explanation of that below.

A Swedish association have blind testet electronic components for many years using what they call before and after testing.

Amplifier Test Method

In their testings, very few amplifiers passed the test and didn't add audible distortion to the listeners. Their testing and my own, have also lead me to believe that the threshold needs to be lower than studies in the past have shown.

Yes, same goes for digital sources and about 20 bits of dynamic range should be transparent even for the most quiet listening rooms.

 

[Atmasphere]

There is, afaik, no evidence that non-blind, non-statistical, individual assessements of poorly-defined percepts, such as audio transparency, have global validity and, yet, I am convinced of my own, as are you by yours.

Transparency is detail (which is to say very faithful to the original signal)- with a lack of annoying distortions, combined with the ability to maintain a proper 3d soundstage. If as a designer, you get these things right in your product, people will say its transparent. If there are gross distortions, if bandwidth or frequency response is screwed up, if there is noise, the circuit will not be perceived as transparent. So if you're wondering how designers get it right so transparency happens, now you have a general map.

And our definition of "fooling" and what is needed to reach it can differ more than our hearing!

I gave you my definition. Every time the system has fooled me my response has been visceral- I was instantly looking for where the sound came from.

Once when I was doing an on-location recording two of my audiophile friends wanted to come along. One was particularly jaundiced IMO- very hard to satisfy (he since has become an orchestra conductor). At one point I realized I wanted to move the mics which were mounted on a single stand. He was listening to the direct mic feed on the headphones. My recording setup was backstage and there were several doors to access it. I went thru the closest one, walked up to the mics and (because they were ribbon mics, very sensitive to motion, so I wanted him to be prepared for what was to follow) I said "OK- I'm going to move the mics now" which was really similar to what I said just before I went through the door. When I got backstage, he was as white as a ghost and looked like he had seen one. The other guy was laughing his ass off. Apparently the sound in the 'phones was so realistic he had thought I was there- instead of having gone through the door. He believed his ears more than his eyes.

That is 'fooling' and is also an example of transparency.

 

[Atmasphere]

Can we know what you consider "our (?) present knowledge about the threshold"? And which little number of amplifiers can be bellow this threshold?

Bjorn and I are on the same page here. If you go over to ASR you'll see people that think that -75-80dB is acceptable, which is IMO where there are so many terrible, bad, really awful bright and harsh solid state amps out there. You really do need the distortion to be either innocuous by nature or so low that it simply will be masked. The problem is that harmonics dictate what instruments sound like. People know this, but oddly seem to rarely make that simple connection. The implication of course is that distortion really really needs to be really low

 

[morricab]

Lower mass=equals lower distortion for mids and highs. This is well known and measurable. It's a big topic where I would encourage you to look at some studies/paper if you want to further dive into the topic. The key is lack and of resonances and the relation to movement and also the cabinet. If you listen to a large wide band driver (for instance 3" to 5") in the highs and compare it to a small tweeter it's very evident despite that the force and movement in the highs are small. But directivity also comes into play.

Distortion is also a matter in the bass and I gave an example of horn loading a woofer, which decreases the distortion compared to front firing. It took four front firing woofers to get close to the pereceived low modulation distortion of two horn loaded 12" woofers in a comparison Klipsch did.
View attachment 101635

Yes horn loaded bass is awesome…my 10 inch horn loaded woofers have so much midbass slam and zero bloat.

 

[microstrip]

Based on the studies and articles I have read over the years and I've lowered certain numbers some to be certain. Explanation of that below.

A Swedish association have blind testet electronic components for many years using what they call before and after testing.

Amplifier Test Method

Sorry, we have no ways of knowing or getting information about "the studies and articles I have read over the years" . The link you provide is not relevant to what we were debating.

In their testings, very few amplifiers passed the test and didn't add audible distortion to the listeners. Their testing and my own, have also lead me to believe that the threshold needs to be lower than studies in the past have shown.

Nice to read your opinion - but sorry I have never read about your tests and you did not provide any links to the swedish group ones.

Yes, same goes for digital sources and about 20 bits of dynamic range should be transparent even for the most quiet listening rooms.

Well, we were not addressing dynamic range, just distortions and noise. According to your criteria, all decent DACs should sound the same.

 

[microstrip]

Bjorn and I are on the same page here. If you go over to ASR you'll see people that think that -75-80dB is acceptable, which is IMO where there are so many terrible, bad, really awful bright and harsh solid state amps out there. You really do need the distortion to be either innocuous by nature or so low that it simply will be masked. The problem is that harmonics dictate what instruments sound like. People know this, but oddly seem to rarely make that simple connection. The implication of course is that distortion really really needs to be really low

IMO you and Bjorn are in very different pages. You focus on distortion that is innocuous by nature and he focuses on extremely low distortions.

However, unfortunately both refer to science and knowledge and do not provide us with proper references or links to such sources. In order to become science something must be clear and accessible to others - secret information about UFOs is not science.

I do not see any point in referring to ASR - they are not an high-end forum.

 

[the sound of Tao]

Yes, but also lot of differences between the individual hearing systems. Rules are created statistically for the large majority - it is why these studies must include a large population.
And unfortunately people often refer to "perceptual rules" but once we ask for them and links to studies documenting them the subject abruptly ends without this "lot" of rules.



Since long OTLs have created the aura of being "transparent" because they do not have an output transformer. It is natural that reviewers focus on this word when reviewing an OTL - this happens in 99% of the OTL reviews I have read. I have owned several OTLs - Futtterman, Technics 20A, Graaf and Atmasphere M50 and M2. I can assure you they sounded quite different from each other, but all reviewers raved about their transparency.

Just to point that IMO reviewers and manufacturers texts can't be a reference on debates about transparency. They are just opinions.


I can't see what words such as "If it measures good and sounds bad, it is bad. If it Measures bad and sounds good, you've measured the wrong thing." add to this debate. But I would love to read good science that can applied to measurements of current SOTA electronics - most I know about only applied to electronics that had much higher distortions and noise than current SOTA.


I have been present to a few performances that I have excellent recordings. It is interesting, but it also carries a risk - when listening to reproduction we focus our attention on the details that we have perceived at the real performance and ignore many others. IMO not a good test for transparency (whichever it means for each of us) . BTW, I find many people just refer to transparency as a synonymous of detailed.



Any system can fool us occasionally on some specific recording. The point is creating a system that fools us most of the time with the music we listen. And our definition of "fooling" and what is needed to reach it can differ more than our hearing!

I’m glad you addressed this Francisco, I struggle with unsubstantiated claims of some mysterious set of absolute human perceptual rules… apparently people without any subject matter expertise like to just make this stuff up.

 

[Bjorn]

Sorry, we have no ways of knowing or getting information about "the studies and articles I have read over the years" . The link you provide is not relevant to what we were debating.

It will take me hours to dig all that up. I'm sure you can find studies on this yourself if you want to. Many of these studies are very old as seen below, and as mentioned I believe the threshold is lower based on newer blind tests.


The 0.1 dB in frequency responso limit comes from Floyd Toole's researchers and are mentioned is his book. On jitter I've read a few papers on different types that were conducted in Japan and the result differed from type of jitter. But no, I don't have time to search for them. nwavguy has written about jitter in the link below but he doesn't link to studies he refers to.

Jitter Does it Matter?

INTRO: Some have asked for more details on jitter. It's a controversial topic and there are lots of myths associated with it. Here’s what I...

nwavguy.blogspot.com

On dynamics, there are some papers here:

AES Journal Forum » Audibility of a CD-Standard A/DA/A Loop Inserted into High-Resolution Audio Playback

The AES Forum is where our members get to interact with each other in between conferences and section meetings.

secure.aes.org

AES E-Library » Dynamic Range Issues in the Modern Digital Audio Environment

The peak sound levels of music performances are combined with the audibility of noise in sound reproduction circumstances to yield a dynamic range criterion for noise-free reproduction of music. This criterion is then examined in light of limitations due to microphones, analog to digital...

www.aes.org

Nice to read your opinion - but sorry I have never read about your tests and you did not provide any links to the swedish group ones.

I can't link to something that's not on the internet. The swedisch association requires a membership and tests are publised in magazines for the subscribers. They have blind testet for many years. What I can tell from these is that the threshold of audibility is lower than several of the old studies we have. That's also my own personal experience when AB testing with level matching. However, I haven't conducted strict ABX tests. So I see no reason to share detailes about these, other than that's my own subjective experience with AB tests.

Well, we were not addressing dynamic range, just distortions and noise. According to your criteria, all decent DACs should sound the same.

You asked whether I believe the same for digital sources. I said yes, and I added what I believe needs to be threshold of dynamic range as well. That's of course on top of the other ones I have mentioned. Meaning few will pass the test. Please read what I'm actually writing.

That being said, I believe THD can be higher than -110 dB to -120 dB and still be transparent. Something around -100/-103 dB here is likely good enough. That's also based on a combination of the swedish blind tests and my own AB tests with level matching, where no difference was perceived when comparing to something that measured better.

Setting an exactly threshold is very challenging. Blind tests can differ and will depend on well they were conducted. Some years ago a radio station blind tested MP3 and found out that 128 kbps was good enough to not sound different from lossless. We have other studies that show that even MP3 in 256 and 320 kbps can be distungished from lossless by trained listeners.

But if we set a very strict threshold for all the types and well below the studies we have, I believe we can be certain.

 

[morricab]

Yes, but also lot of differences between the individual hearing systems. Rules are created statistically for the large majority - it is why these studies must include a large population.
And unfortunately people often refer to "perceptual rules" but once we ask for them and links to studies documenting them the subject abruptly ends without this "lot" of rules.



Since long OTLs have created the aura of being "transparent" because they do not have an output transformer. It is natural that reviewers focus on this word when reviewing an OTL - this happens in 99% of the OTL reviews I have read. I have owned several OTLs - Futtterman, Technics 20A, Graaf and Atmasphere M50 and M2. I can assure you they sounded quite different from each other, but all reviewers raved about their transparency.

Just to point that IMO reviewers and manufacturers texts can't be a reference on debates about transparency. They are just opinions.


I can't see what words such as "If it measures good and sounds bad, it is bad. If it Measures bad and sounds good, you've measured the wrong thing." add to this debate. But I would love to read good science that can applied to measurements of current SOTA electronics - most I know about only applied to electronics that had much higher distortions and noise than current SOTA.


I have been present to a few performances that I have excellent recordings. It is interesting, but it also carries a risk - when listening to reproduction we focus our attention on the details that we have perceived at the real performance and ignore many others. IMO not a good test for transparency (whichever it means for each of us) . BTW, I find many people just refer to transparency as a synonymous of detailed.



Any system can fool us occasionally on some specific recording. The point is creating a system that fools us most of the time with the music we listen. And our definition of "fooling" and what is needed to reach it can differ more than our hearing!

With regard to the science Micro. You know very well that the best one can do, scientifically, to relate measurements to perception is a statistical correlation, which may or may not lead to causation. This is true for every aspect of human psychology and since what we hear in actuality is only in our heads as a response to stimulus then audio is all about psychology.

Designers should strive for achieving a good model for human hearing and then designed gun their products accordingly, assuming they want the widest overall appeal and not just what sounds really that to them. This would be a science based design approach that STILL would not appeal to everyone because of the inherent “fuzziness” in human psychology.

Maybe if they achieved inherently distortionless device that truly made no distortion that would end the debate…but probably not and it is far from ever happening.

Studies have shown that there is no correlation between THD/IMD and perception of distortion…so anyone touting ultralow THD is simply ignorant of work that has been done regarding this metric. Ralph over simplifies to the point of being wrong.

i will have to say though that my experience with OTLs is that they do sound perceptibly more transparent than other tech but some sound too lean tonally. Not all though and those are amps no the best amps ever. The bigger problem was always reliability and heat heat heat.

 

[microstrip]

It will take me hours to dig all that up. I'm sure you can find studies on this yourself if you want to. Many of these studies are very old as seen below, and as mentioned I believe the threshold is lower based on newer blind tests.
View attachment 101668

The 0.1 dB in frequency responso limit comes from Floyd Toole's researchers and are mentioned is his book. On jitter I've read a few papers on different types that were conducted in Japan and the result differed from type of jitter. But no, I don't have time to search for them. nwavguy has written about jitter in the link below but he doesn't link to studies he refers to.

Jitter Does it Matter?

INTRO: Some have asked for more details on jitter. It's a controversial topic and there are lots of myths associated with it. Here’s what I...

nwavguy.blogspot.com

On dynamics, there are some papers here:

AES Journal Forum » Audibility of a CD-Standard A/DA/A Loop Inserted into High-Resolution Audio Playback

The AES Forum is where our members get to interact with each other in between conferences and section meetings.

secure.aes.org

AES E-Library » Dynamic Range Issues in the Modern Digital Audio Environment

The peak sound levels of music performances are combined with the audibility of noise in sound reproduction circumstances to yield a dynamic range criterion for noise-free reproduction of music. This criterion is then examined in light of limitations due to microphones, analog to digital...

www.aes.org

I can't link to something that's not on the internet. The swedisch association requires a membership and tests are publised in magazines for the subscribers. They have blind testet for many years. What I can tell from these is that the threshold of audibility is lower than several of the old studies we have. That's also my own personal experience when AB testing with level matching. However, I haven't conducted strict ABX tests. So I see no reason to share detailes about these, other than that's my own subjective experience with AB tests.


You asked whether I believe the same for digital sources. I said yes, and I added what I believe needs to be threshold of dynamic range as well. That's of course on top of the other ones I have mentioned. Meaning few will pass the test. Please read what I'm actually writing.

That being said, I believe THD can be higher than -110 dB to -120 dB and still be transparent. Something around -100/-103 dB here is likely good enough. That's also based on a combination of the swedish blind tests and my own AB tests with level matching, where no difference was perceived when comparing to something that measured better.

Setting an exactly threshold is very challenging. Blind tests can differ and will depend on well they were conducted. Some years ago a radio station blind tested MP3 and found out that 128 kbps was good enough to not sound different from lossless. We have other studies that show that even MP3 in 256 and 320 kbps can be distungished from lossless by trained listeners.

But if we set a very strict threshold for all the types and well below the studies we have, I believe we can be certain.

Thanks for you time, but a cut and paste from an interesting 2005 Audioholics article https://www.audioholics.com/room-acoustics/human-hearing-distortion-audibility-part-3 adds nothing new to the subject of our discussion - THD thresholds for transparency. And adding new subjects, such as dynamics or frequency response just brings noise to the main subject.

My point is simple - existing audio science can't bring quantitative data to our discussions on transparency in current high-end stereo sound reproduction. People can have their beliefs based on their experience and report them, we are happy to read and debate them, but they are no way scientific.

Surely when you have data that shows the contrary I will be happy to change my mind.

 

[pjwd]

Thanks for you time, but a cut and paste from an interesting 2005 Audioholics article https://www.audioholics.com/room-acoustics/human-hearing-distortion-audibility-part-3 adds nothing new to the subject of our discussion - THD thresholds for transparency. And adding new subjects, such as dynamics or frequency response just brings noise to the main subject.

My point is simple - existing audio science can't bring quantitative data to our discussions on transparency in current high-end stereo sound reproduction. People can have their beliefs based on their experience and report them, we are happy to read and debate them, but they are no way scientific.

Surely when you have data that shows the contrary I will be happy to change my mind.

Change your mind about what ?

 

[microstrip]

Change your mind about what ?

On the relevance of audio science in the correlation of THD thresholds with transparency in high-end electronics.