Hi mayhem13,
I hear you. However, a specified room and speaker positioning is probably not going to happen. The home would have to be designed for this specifically. The space outside this area will also affect the low bass. It would be nice though.
I don't know about the number of required speakers. I do know that you are multiplying the cost over a stereo system. Just seeing a good 5 channel Dolby surround system is very rare. Add the electronics at that same two channel quality level, and having the decoder do it's job with low distortion is a challenge again. It could be done, but that system is simply out of reach for most people. They struggle to get a decent two channel system at any price. Then, there is the ever present WAF (luckily I didn't have a problem with mine, but it was still a factor).
I hear you. However, a specified room and speaker positioning is probably not going to happen. The home would have to be designed for this specifically. The space outside this area will also affect the low bass. It would be nice though.
I don't know about the number of required speakers. I do know that you are multiplying the cost over a stereo system. Just seeing a good 5 channel Dolby surround system is very rare. Add the electronics at that same two channel quality level, and having the decoder do it's job with low distortion is a challenge again. It could be done, but that system is simply out of reach for most people. They struggle to get a decent two channel system at any price. Then, there is the ever present WAF (luckily I didn't have a problem with mine, but it was still a factor).
I picked that up from my brief read earlier, but that is not what happens in reality. In reality, hierarchical models learned from experience process objects (aka data structures) in parallel, and the most pressing surprise (at whatever hierarchical level it occurs) is what leads to a new perception. Neural resources are not "allocated", hence why I believe the parallel Bayesian model would be more fruitful.
Hmm, very hard to understand what you say without education for the subject 
could you link to any papers, or other material on what you refer to? thanks!
Example, what I'm thinking writing all the above, is change in perception I can make happen taking single step forward closer to speakers and suddenly clarity and envelopment happens, as if stepping inside the sound, or take a step back out from it and now the sound is more hazy and in front. Like passage between my room and the sound of the recording. Not sure if this is something that makes a pressing surprise? ai can make this at will, repeatedly day in day out.
could you link to any papers, or other material on what you refer to? thanks!Example, what I'm thinking writing all the above, is change in perception I can make happen taking single step forward closer to speakers and suddenly clarity and envelopment happens, as if stepping inside the sound, or take a step back out from it and now the sound is more hazy and in front. Like passage between my room and the sound of the recording. Not sure if this is something that makes a pressing surprise? ai can make this at will, repeatedly day in day out.
Last edited:
WRT your string section, most of the requisite information can be ascertained from the cross-bispectrum and bispectrum that relate (or not) different frequencies together, where the former allows for spatial discrimination too. It is highly likely these tools model very well the processes apparent in our auditory processing. Indeed, you would need to find a very good metabolic argument as to why the brain did it in a more complex manner.
As a further point, I have also been lucky enough to be seated in various auditoria for live performances that were being recorded. On every occasion, hearing afterwards the recording in the engineer's seat was significantly more revealing of detail and better balanced than the actual performance I heard in my seat. I learned to trust the engineers to produce the better benchmark, and I suggest that retrieving the information contained in their recording is how the accuracy of loudspeakers are best judged.
Probably the best non-mathematical introductory read might be "On Intelligence" by Hawkins and Blakeslee? The book describes an attempt to reverse engineer the cerebrum and is told from an engineering perspective too (rather than a neurological one).Hmm, very hard to understand what you say without education for the subject
Clearly you’ve lost your understanding of both the creation of and consuming of music is an art…..art is and always will be subjective. To discredit my creative intent based on the assumptions of predicted and applied physics speaks for itself…..you’ve lost all passion for the art…..sorry. Majority listening environments and gear has little to nothing in relation to high fidelity.…..the two are still diverging today and advances in tech will only cater to more flexible, portable and accessible solutions. With large LCD screens of 85” under a grand, the platform for what I suggest now exists and the sound bar market is booming beyond any and all AV tech today…….exclude the ‘high fidelity’ label which sound bars simply cannot fulfill in their current LCR format or ambiophonic nonsense. 7 identical two way discreet systems in a horizontal array…..80” long and a single mono subwoofer channel or stereo subs?……that’s a sub $1500 market piece of gear and less entry level with full range drivers…….all that’s needed is the vision for development…..the algorithm is easy……and for Dolby, the 4 unused channels could easily be repurposed for sound field enhancement or object steering. Immersive audio and Atmos audio is a joke…..nobody in the market is cutting holes in their ceilings or surrounding themselves with satellite speakers.
That and if artists ever congregate and decide that fractional pennies per stream revenue isn’t cutting it, here’s there opportunity to regain a market share for physical media where such a system would only be compatible with a purchase option.…..
Hi mayhem13,
It is art during the creation process. Once it is packaged, it is a product. You are not the artist. I think you've completely lost the concept.
If you want to buy a picture and paint it, go ahead. That's your prerogative. If you want to buy music in some form and run it through a meat grinder of a system, again that is your deal alone. But the rest of the world paid for a product, and the skill of the engineers, producers and studio techs. It would be pretty stupid to redefine what they created since we paid for their skill.
Reproduction is not an art. It is an evolving science that steadily improves. The better it gets, the more people like it (not terribly surprising). Pirate recordings were fun (remember those?). They were pretty raw, rough and sounded good on substandard systems when we were young. But given a nicely produced and engineered source, who doesn't want the best from it?
It is art during the creation process. Once it is packaged, it is a product. You are not the artist. I think you've completely lost the concept.
If you want to buy a picture and paint it, go ahead. That's your prerogative. If you want to buy music in some form and run it through a meat grinder of a system, again that is your deal alone. But the rest of the world paid for a product, and the skill of the engineers, producers and studio techs. It would be pretty stupid to redefine what they created since we paid for their skill.
Reproduction is not an art. It is an evolving science that steadily improves. The better it gets, the more people like it (not terribly surprising). Pirate recordings were fun (remember those?). They were pretty raw, rough and sounded good on substandard systems when we were young. But given a nicely produced and engineered source, who doesn't want the best from it?
Making accurate loudspeakers is an engineering function. Making music is an art. I am not confused by the difference, and to describe me as having "lost all passion for art" is as ridiculous as it is presumptuous. It is also personal, not factual, and should not be posted here. The physics stands by itself.
@soundbloke , at some point in this discussion you encouraged someone to listen to the leading edge of a piano note on an ESL as compared to a cone or dome speaker. Would you be willing to expound on the differences (I have had ESLs in the distant past, but not currently, so can't make the comparison myself) you would expect? We have all read the possible benefits/limitations of ESLs; are you able to clarify any of these or compare/contrast them as they relate to some of the deep topics discussed in this thread?
Thank you,
Bill
Thank you,
Bill
Can you explain how the "cross-bispectrum and bispectrum" provide more info than the usual CCIR & SMPT intermod tests?
I've been involved in practically ALL da various fancy displays; from the earliest BBC 'delayed resonances' to cepstrum & wavelet buzzwords. They ALL show the same info; just display it in different colours
Laurie Fincham, who pioneered CDS (da waterfall ubiquitous in all modern speaker measurement packages, which we called KEFplots) claimed he could tell important stuff just by looking at the IR
but I would take that with a big pinch of salt There were also PAFplots (after our Peter A Fryer) and JAPplots (from a Japanese company I can't remember). Later on we were Beta testers for various B&K gear involving cepstrums, wavelets and other buzzwords.
The motivation for that comment was that ESLs have no distortion mechanism subjectively commensurate with the eddy-related non-linearities in moving coil drivers. I suggest making the comparison between the two would enable listeners to more readily identify the eddy-related non-linearities that have been discussed in this thread. That is not to say that I believe the ESLs will necessarily deliver a result that is more accurate in all aspects of its performance, however.
Accurate to what?…..what’s the reference?….physics, biology, a particular individual or group of engineers?….a prescribed environment?…..too many holes in your argument.
mayhem13,
What is your reference?
You seem to simply wish to muddy the waters. I don't see any holes at all in his point of view. It's easy to be the opposition without having to stand on anything solid.
So the thread question has to do with measurements and whether they are useful or not. What is your position? Why is that your position and can you defend your position effectively with actual knowledge? From what you have said so far, I don't think you can do that.
What is your reference?
You seem to simply wish to muddy the waters. I don't see any holes at all in his point of view. It's easy to be the opposition without having to stand on anything solid.
So the thread question has to do with measurements and whether they are useful or not. What is your position? Why is that your position and can you defend your position effectively with actual knowledge? From what you have said so far, I don't think you can do that.
Thank you. So not willing to flesh out a bit the pros and cons, or perhaps which technology has the best potential for the highest fidelity after weighing the pros and cons? Or maybe measurements that are important that we can look at that would explain some of the audible differences we perceive?
Sorry to push,
Bill
I will answer briefly now and follow it up when I have more time if required...Can you explain how the "cross-bispectrum and bispectrum" provide more info than the usual CCIR & SMPT intermod tests?
I've been involved in practically ALL da various fancy displays; from the earliest BBC 'delayed resonances' to cepstrum & wavelet buzzwords. They ALL show the same info; just display it in different colours
Laurie Fincham, who pioneered CDS (da waterfall ubiquitous in all modern speaker measurement packages, which we called KEFplots) claimed he could tell important stuff just by looking at the IR
There were also PAFplots (after our Peter A Fryer) and JAPplots (from a Japanese company I can't remember). Later on we were Beta testers for various B&K gear involving cepstrums, wavelets and other buzzwords.
The Cepstrum and other like spectra are second order analyses. The Cepstrum is useful for establishing excess phase components in a linear system, for example. But it is still essentially a spectrum where we can illustrate how a single frequency will respond - and if we are to infer a subjective response, how we will respond to that single frequency at our ears.
The waterfall plots afford us the opportunity to display much the same information in time and frequency simultaneoulsy, but are still nevertheless telling us information about a single frequency. Notably re post #24, all cumulative spectra are smeared Wigner Distributions and therein lies an opportunity to improve what we can subjectively assess from such displays too.
But the bispectra are third order analyses and enable one to see how the response to one frequency is related to another, for example, how harmonics relate to a fundamental. Essentially we can group information together (or not as may also be the case) via such processing in order to separate information that is not discernible in second order spectral analyses.
Perhaps one way to think of things is that energy spectra do away with phase information, and the bispectra add it back in. We might find here, for example, the means to model how we perceive transient information differently to steady state information, how we tell voices apart in a harmony, how we characterise one louspeaker from another, for examples. Via the cross-bispectrum enabled by having two ears or two microphones, we have the additional means to model how we discern different sound sources in space, and how we discern a sound source from reflections in a room, for examples.
So the bispectra are not a new panacea or a replacement for other measurements. Instead they are a valuable addition in retrieving information that is not apparent via second order analyses and can help in subjective modelling (if not actually replicate the processing that occurs in our brains). As previously commented, however, displaying bispectra in an easily "read" format is a challenging undertaking in its own right.
I am sufficiently experienced not to trust myself to make such judgements.
The particular non-linearities I referred to can be discerned via suitable measurements and displays. I think I reported earlier in this thread (?) that the rise in distortion at mid-frequencies is one characteristic (and notably different to the fall with frequency associated with displacement dependent non-linearities). The distortion spectrum is also often spiky, although I am not clever enough to relate the spikes to the motor geometry differences that appear to cause them.
But how it sounds to you is best left for you to decide. And that will be influenced by all sorts of other factors too - source material, other distortions, how good a day you have had and so on. Given this is a DIY forum, I wholeheartedly encourage people to experiment for themselves and not to trust my opinion.
As a starting point, accurate refers to the fidelity in recreating the sound field or sound pressures at the listener's ears that would have been apparent had the listener been sat at the microphone position in the original recording environment.
The only proviso is that it excludes material generated in a studio where there is no acoustic reference, or material recorded in an acoustic environment by means that prevent the above definition being physically realisable. Here, as I clearly suggested in a previous post, accuracy is not easily definable.
No holes, no argument.
Unfortunately, no. I am generally not good at explaining things.
1) learn the art and science of general acoustic measurements. Tutorials on B&K site are very helpful. Do not expect quick progress. There are so many mistakes you can do ... I've done them all, several times, over and over again.
2) read the .pdf I provided with .m code.
3) equalize condenser and omni measurement mic for the chosen setup
4) start with doc_pa02.m script. Modify the first part the way you want it, choose the levels, measurement method, source, coloration, and options of pass-band limiting, and see what comes out.
5) examine that everything in the graphs and recordings makes (at least some) sense. It surely won't, so you'll need to figure out why. Play with it.
Do not hesitate to ask me any specific questions you may have, suggest the features you need, etc.