First order crossovers are theoretically phase neutral, if you can blend in the tweeter and woofer using just a single cap on the tweeter and a coil on the woofer.
In practicality, this doesn't work very well for many speakers because first-order crossovers don't cross out with enough slope to avoid problems. So many designers prefer to have a few phase issues in a smaller crossover band with 2nd order and higher networks.
But I should point out that if you are trying to preserve the "characteristic" sound of a driver, this is barking up the wrong tree from my design philosophy.
You don't want speakers to have "characteristic". They should be neutral. The box should be made of solid, non-resonant material and braced if necessary, because any vibrations of the box, no matter what the material is, is unwanted noise. This is one of the main reasons why hi-density particle board (MDF) is actually better than plywood, even though plywood is stronger. Plywood is more resonant. MDF is more stiff (even if it is less strong) and thus makes a better speaker.
You can put one of those cheap fisher 12inch paper drivers that were so common in the 80s (in a wide variety of japanese and american consumer systems) into a good, solid braced box. You would be amazed at how much cleaner it sounds. Most of these drivers were fitted into boxes made of 1/2-inch cheap grade particle board, with undersized box volume to give a rise in the midbass for fake bass that helps the speaker sell in the showroom. Most of the sound below 200 hz. on these crappy speakers is box noise, rather than the actual sound coming from the woofer. In a good box, the sound is surprisingly clean.
In practicality, this doesn't work very well for many speakers because first-order crossovers don't cross out with enough slope to avoid problems. So many designers prefer to have a few phase issues in a smaller crossover band with 2nd order and higher networks.
But I should point out that if you are trying to preserve the "characteristic" sound of a driver, this is barking up the wrong tree from my design philosophy.
You don't want speakers to have "characteristic". They should be neutral. The box should be made of solid, non-resonant material and braced if necessary, because any vibrations of the box, no matter what the material is, is unwanted noise. This is one of the main reasons why hi-density particle board (MDF) is actually better than plywood, even though plywood is stronger. Plywood is more resonant. MDF is more stiff (even if it is less strong) and thus makes a better speaker.
You can put one of those cheap fisher 12inch paper drivers that were so common in the 80s (in a wide variety of japanese and american consumer systems) into a good, solid braced box. You would be amazed at how much cleaner it sounds. Most of these drivers were fitted into boxes made of 1/2-inch cheap grade particle board, with undersized box volume to give a rise in the midbass for fake bass that helps the speaker sell in the showroom. Most of the sound below 200 hz. on these crappy speakers is box noise, rather than the actual sound coming from the woofer. In a good box, the sound is surprisingly clean.
Wow this thread really took off 🙂 Sorry to have not replied for so long, and apologies to anyone who took offense to the confrontational tone of my initial postings. Knowing that speaker design is all about tradeoffs, I was interested in what attributes the designer of this speaker 'traded-off' for, since clearly they did not focus flat on/off axis frequency or resonance control. I was also interested in how to measure how well the speakers did at these other attributes the designer focued on. Several posters have addressed both issues on this thread, my thanks to them. There's a lot of information to absorb, and I'll post more when I have.
Regarding measurements telling us how a speakers sounds - they do. Of course just a FR curve doenst tell you that much, but comine that with off axis FR curves, waterfall plots, impedance plots, phase plots, and distortion tests, and you have a lot of information about how the speaker performs. There will never be one measure that tells us how a speaker sounds, but I think with a whole suite of measurements and proper interpretation (obviously my weak point, and what I made this thread to improve on) you can get a good idea.
As for the argument 'some people like non-flat FR'. Of course thats true. But if we can agree for a moment that accuracy is the goal in designing speakers, then while some may PREFER the coloured sound, it is not a good (read: accurate) transducer. The same can be said of distortion - some prefer their bass with lots of distortion to 'fatten' the sound a bit. Thats good for them, and I won't second guess their choice. But it is a choice away from accuracy, which is my goal in a speaker. Furthermore, the argument 'you'll never have perfect FR, so don't bother' is IMO self-defeating for a speaker builder. By the same rationale, you can never have zero distortion, so why try to get it as low as possible? Or perfect transient response, so why bother trying to get it as close as possible? Or for that matter, a perfectly effecient engine, so why try to improve fuel economy.
While its obvious we can't all agree on what sounds good, I would hope we can all agree on attributes of an accurate transducer (low distortion, flat frequency response, good transient response, etc.) These are 2 separate issues that are being used interchangeably when they are not at all. So I guess a better question than my initial one would be: The designer of this speaker obviously tried to optimize the design to maximize certain attributes. Which attributes did they design for that pertain to accuracy (I'm guessing the mentioned phase coherency for a start) and which ones pertain to a 'signature sound' (ie intentional deviations from flat FR to produce a 'coloured' yet desireable sound). I am much more interested in the design decisions take and ways in which we (and the designer) can evaluate how successfull these decisions were (including measurements).
Regarding measurements telling us how a speakers sounds - they do. Of course just a FR curve doenst tell you that much, but comine that with off axis FR curves, waterfall plots, impedance plots, phase plots, and distortion tests, and you have a lot of information about how the speaker performs. There will never be one measure that tells us how a speaker sounds, but I think with a whole suite of measurements and proper interpretation (obviously my weak point, and what I made this thread to improve on) you can get a good idea.
As for the argument 'some people like non-flat FR'. Of course thats true. But if we can agree for a moment that accuracy is the goal in designing speakers, then while some may PREFER the coloured sound, it is not a good (read: accurate) transducer. The same can be said of distortion - some prefer their bass with lots of distortion to 'fatten' the sound a bit. Thats good for them, and I won't second guess their choice. But it is a choice away from accuracy, which is my goal in a speaker. Furthermore, the argument 'you'll never have perfect FR, so don't bother' is IMO self-defeating for a speaker builder. By the same rationale, you can never have zero distortion, so why try to get it as low as possible? Or perfect transient response, so why bother trying to get it as close as possible? Or for that matter, a perfectly effecient engine, so why try to improve fuel economy.
While its obvious we can't all agree on what sounds good, I would hope we can all agree on attributes of an accurate transducer (low distortion, flat frequency response, good transient response, etc.) These are 2 separate issues that are being used interchangeably when they are not at all. So I guess a better question than my initial one would be: The designer of this speaker obviously tried to optimize the design to maximize certain attributes. Which attributes did they design for that pertain to accuracy (I'm guessing the mentioned phase coherency for a start) and which ones pertain to a 'signature sound' (ie intentional deviations from flat FR to produce a 'coloured' yet desireable sound). I am much more interested in the design decisions take and ways in which we (and the designer) can evaluate how successfull these decisions were (including measurements).
Phase coherancy
Coherant to what? The other drivers? I suppose. So they are all phase aligned with eachother. This beg the question of course: what about the other speakers in the stereo (or x.1) system?
I will wager this is not what makes them sound good - reason being that so much depends on where you stand - I mean, sit. (Chruchill said that where you stand depends on where you sit!).
I think the reason they sound good is because the cone moves in proportion to the wave form driving it in amplitude terms. So we get into damping, compliance, mass, cone resonance and so on.
I'm still looking for metrics I believe. "Distortion" per se, is not it because one can characterize any failure to analog the input waveform as distortion. So, while Distortion may capture the idea, there's no new content in the concept. There's many kinds of departure from the input. Even departure from flat is a sort of distortion. My basic inquiry here (and it IS a good thread) is:
What are the variables and what are the coeffecients? It must be possible to at least characterize good sounding speakers. (Is there a Bose conspiracy to keep such a measure out of the marketplace?). Almost anyone with an ear will tell you, and get agreement, that speaker A sounds better than B, if it does. In the land of esoterica intellegent ears disagree - but that's esoterica.
Any fool will AB a set of Bose with a set of (good speakers) and tell you the good speakers sound better. The room makes a difference, but I bet the same casual observer could tell the difference in an anechoic chamber. My wife, who knows nothing about audio, can tell my successses from my failures in speaker building. Why can't we measure it? and, if we can.... what are the variables, and what are the coeffecients, or how would we determine them?
Coherant to what? The other drivers? I suppose. So they are all phase aligned with eachother. This beg the question of course: what about the other speakers in the stereo (or x.1) system?
I will wager this is not what makes them sound good - reason being that so much depends on where you stand - I mean, sit. (Chruchill said that where you stand depends on where you sit!).
I think the reason they sound good is because the cone moves in proportion to the wave form driving it in amplitude terms. So we get into damping, compliance, mass, cone resonance and so on.
I'm still looking for metrics I believe. "Distortion" per se, is not it because one can characterize any failure to analog the input waveform as distortion. So, while Distortion may capture the idea, there's no new content in the concept. There's many kinds of departure from the input. Even departure from flat is a sort of distortion. My basic inquiry here (and it IS a good thread) is:
What are the variables and what are the coeffecients? It must be possible to at least characterize good sounding speakers. (Is there a Bose conspiracy to keep such a measure out of the marketplace?). Almost anyone with an ear will tell you, and get agreement, that speaker A sounds better than B, if it does. In the land of esoterica intellegent ears disagree - but that's esoterica.
Any fool will AB a set of Bose with a set of (good speakers) and tell you the good speakers sound better. The room makes a difference, but I bet the same casual observer could tell the difference in an anechoic chamber. My wife, who knows nothing about audio, can tell my successses from my failures in speaker building. Why can't we measure it? and, if we can.... what are the variables, and what are the coeffecients, or how would we determine them?
well certainly a metric would be nice. But I think it would have to be an index or series of indexes. That is, it would have to take several measurements that are illustrative of a given attribute, for example measurements of parameters that affect imaging would be grouped together for an 'imaging index', other for a 'transient response' index. Then these indexes could be combined into an 'overall index'. The hard part of course would be getting people to agree on what measures are indicative of these attributes (ie does flat power response contribute to imaging? What about impulse response?).
Yes!
Now we're getting somewhere.
I would be happy with a series of indexes. Then you could argue about the weights among them - but everyone would agree that if the transient index were below some threshold regardless of all other indexes, they would sound like crap. Right? We can all argue about the grey area, but "consumers" would be better served by any bit of clarity. I went into a Cambrige Audio store this weekend in Boston and asked about the specs on their top speaker. They said: We don't publish specs because we think they lie. ... and our drivers are made in China.
So, its come to that. Even a somewhat respectable audio store thinks there's no good measure so consumers are forced to go with their ears (good) - but with no real comparison.
How would that be for cars? We don't publish horsepower - or mpg - we feel the specs are misleading. So, the car salesman ends up with more to back up his or her case than the audio salesman. Is audio like perfume? No! Is it the opinion of this group that idiots might as well live with their ignorance? If we could find the right spec then quality would win, crap would lose, and the price of quality would go down for everyone.
Now we're getting somewhere.
I would be happy with a series of indexes. Then you could argue about the weights among them - but everyone would agree that if the transient index were below some threshold regardless of all other indexes, they would sound like crap. Right? We can all argue about the grey area, but "consumers" would be better served by any bit of clarity. I went into a Cambrige Audio store this weekend in Boston and asked about the specs on their top speaker. They said: We don't publish specs because we think they lie. ... and our drivers are made in China.
So, its come to that. Even a somewhat respectable audio store thinks there's no good measure so consumers are forced to go with their ears (good) - but with no real comparison.
How would that be for cars? We don't publish horsepower - or mpg - we feel the specs are misleading. So, the car salesman ends up with more to back up his or her case than the audio salesman. Is audio like perfume? No! Is it the opinion of this group that idiots might as well live with their ignorance? If we could find the right spec then quality would win, crap would lose, and the price of quality would go down for everyone.
The main problem is that some of those parameters have to be traded against each other. As I already mentioned: It is definitely not easy to achieve good temporal response, flat amplitude response and low THD/IMD at the same time. The fist one requires low order crossovers and single-driver or two-way constructions (though more ways are possible but less common in this types). The second and third criteria are best matched with steep crossovers and three- (or more- ) way designs.
The speaker discussed here has one of the best step responses I have seen so far (I am talking about the initial transient and NOT the 1 kHz ringing which seems to come from the reflex tunnel acting as a TML).
There are in fact people who propose a measurement of stereo imaging, John Watkinson for example. He is also proposing an additional measure for accuracy: Information capacity. They performed tests with compressed and uncompressed sounds. The bitrate that was undiscernible from uncompressed digital audio, when reproduced through the speaker under test, was that speaker's information capacity.
Some more info can be found under:
http://www.celticaudio.co.uk/articles/science.pdf
Regards
Charles
I had an idea to test speakers against one another in localization properties. I started with the idea that a good loudspeaker when played by itself should be easily localized. I'm talking about a full range speaker, 2-way, or 3-way, or whatever you might use where frequency response is present above the bass region all the way to 20k. I figure that the ideal speaker would naturally preserve all of the sounds used by the ear to localize. If used by itself, mono, then what you would be localizing would be the speaker itself. Therefore if one wanted to test the ability of a speaker to preserve localization cues then a simple test could be done.
The air force has a sound lab containing a geodesik sphere that has loudspeakers at each of its many many verticies. The speakers all have 3 LED's above them (two green and one red if I remember correctly). Using this sphere they can do several localization tests. One test that could be done using this sphere might be a test that truely rates a loudspeaker's ability to play localization artifacts. Putting a listener in the center of the sphere we would replace a single speaker with the test speaker. Then playing a localizable sound through it we would measure the amount of time taken to localize the speaker. They use those little red and green lights for this, the speaker with only a green light showing is playing or something like that. So once the listener finds the green light only speaker, he has localized the sound. The test would be repeated mant times, moving the speaker to different places. An average time would be found for that speaker and listener combination. Then the next test speaker would be used. The same thing would be done. The final result would be, whichever has a shorter average time. Was localized more easily, therefore preserved more localization cues.
The air force has a sound lab containing a geodesik sphere that has loudspeakers at each of its many many verticies. The speakers all have 3 LED's above them (two green and one red if I remember correctly). Using this sphere they can do several localization tests. One test that could be done using this sphere might be a test that truely rates a loudspeaker's ability to play localization artifacts. Putting a listener in the center of the sphere we would replace a single speaker with the test speaker. Then playing a localizable sound through it we would measure the amount of time taken to localize the speaker. They use those little red and green lights for this, the speaker with only a green light showing is playing or something like that. So once the listener finds the green light only speaker, he has localized the sound. The test would be repeated mant times, moving the speaker to different places. An average time would be found for that speaker and listener combination. Then the next test speaker would be used. The same thing would be done. The final result would be, whichever has a shorter average time. Was localized more easily, therefore preserved more localization cues.
The problem is that we don't want to localise (translation into American: localize) the speakers but the "phantom sources" generated by stereo reproduction.
With stereo reproduction you don't want the speakers to be localised easily. The only situuation when you would want this is if you have content that is panned to the extreme left or right.
Regards
Charles
With stereo reproduction you don't want the speakers to be localised easily. The only situuation when you would want this is if you have content that is panned to the extreme left or right.
Regards
Charles
Chances are that added mass to the center of the woofer is the main method used to mechanically cross out the driver.
If so, the speaker's response to transients would actually be worse than if it had a simple crossover even though there are no capacitors or coils.
The nice thing about an air core inductors is that the crossover slope is much more precisely controlled than with this mechanical rolloff, so you can make the speaker much more accurate.
I really doubt that the tradeoffs necessary to do a mechanical rolloff of the woofer don't cause a degradation in transient behavior.
Given the lack of accuracy of this speaker with frequency response, there would be a lot of irritating colorations of the sound. The curve is so bad that you would occassionally find a musical passage to be rendered with some notes much louder or quieter than they should be. A lot of the "music" would be lost on some recordings.
If so, the speaker's response to transients would actually be worse than if it had a simple crossover even though there are no capacitors or coils.
The nice thing about an air core inductors is that the crossover slope is much more precisely controlled than with this mechanical rolloff, so you can make the speaker much more accurate.
I really doubt that the tradeoffs necessary to do a mechanical rolloff of the woofer don't cause a degradation in transient behavior.
Given the lack of accuracy of this speaker with frequency response, there would be a lot of irritating colorations of the sound. The curve is so bad that you would occassionally find a musical passage to be rendered with some notes much louder or quieter than they should be. A lot of the "music" would be lost on some recordings.
The paper about the testing is interesting, but why are there no test results?
It would be interesting to rate some well-known speakers with this method and see if the results correlate with experience.
It would be interesting to rate some well-known speakers with this method and see if the results correlate with experience.
1st rule of loudspeakers - listen, if you like them and can be bothered, look at the measurements, if they look bad, ignore them and go with your ears. The only really important graph is that for impedence, will it wreck your amplifier or not; even that is moot, as it is a single frequency measurement it is only really relevant when running test tones, isn't it? If you are sending a whole range of frequencies at the same time, is the single curve an accurate representation of the actual run-time impedence, shouldn't the test be done playing several frequencies at once?
I agree with you that ears matter over anything else. I have not had a chance to hear this one, so perhaps I shouldn't be throwing stones yet.
I note that earlier in the thread a mention was made of the lack of harmonic, and presumably, intermodulation, distortion measurements was made with regard to speakers. As soon as someone starts measuring tubas, clarinets, violins and such for their distortion paramaters I'll believe that they are relevant. Given that a pipe of length X produces, when excited, a note Y, it is the harmonics and intermodulations that distinguish a clarinet playing a "C#" from a Cor Anglais and an Oboe generating the same note.
This may explain why vinyl sounds more real than CDs or DAB; music is an analogue process, it is continuous with time, whereas digital reproduction is contiguous, in that the constituent parts lie alongside each other but do not touch. Given any waveform, if you slice it with equal mark/space ratios, no matter how many thousand times you slice it, you are still discarding 50% of the music, to dither about at a later stage to try filling in the gaps thereby created is only going to compromise the music.
I think that it is now accepted that amplifiers require a bandwidth running up to as much as 200,000hz to accurately reproduce the, unheard, harmonics of the music, so is sampling at 90Khz/180Khz adequate?
Imagine you go to your local bakers and ask for a multigrain brown loaf. He hands you a pack of sliced bread which weighs half what you expect it to; hang on you say, I didn't ask for a sliced loaf, and it doesn't feel right. Ah well sir, we only sell digital bread, its the same length as an old loaf, but you have one thousand 0.5mm sliceswith 0.5mm gaps in between, I'm sure sir will think it tastes the same as sir's old loaf. To me the difference between musical instruments is like the distribution of the grain within the bread, it is what gives each loaf its distinctive taste; and slice it whichever way you will, you are still removing half the grain.
This may explain why vinyl sounds more real than CDs or DAB; music is an analogue process, it is continuous with time, whereas digital reproduction is contiguous, in that the constituent parts lie alongside each other but do not touch. Given any waveform, if you slice it with equal mark/space ratios, no matter how many thousand times you slice it, you are still discarding 50% of the music, to dither about at a later stage to try filling in the gaps thereby created is only going to compromise the music.
I think that it is now accepted that amplifiers require a bandwidth running up to as much as 200,000hz to accurately reproduce the, unheard, harmonics of the music, so is sampling at 90Khz/180Khz adequate?
Imagine you go to your local bakers and ask for a multigrain brown loaf. He hands you a pack of sliced bread which weighs half what you expect it to; hang on you say, I didn't ask for a sliced loaf, and it doesn't feel right. Ah well sir, we only sell digital bread, its the same length as an old loaf, but you have one thousand 0.5mm sliceswith 0.5mm gaps in between, I'm sure sir will think it tastes the same as sir's old loaf. To me the difference between musical instruments is like the distribution of the grain within the bread, it is what gives each loaf its distinctive taste; and slice it whichever way you will, you are still removing half the grain.
Digital will eventually sound just as good as analog. If they make the time slice small enough, the artifacts won't matter.
I do agree that there is a slight edge to digitally recorded music at the 44 kHz sampling rate. I have not heard anything sampled at a higher frequency to know if this eliminates the issues. But even at 44khz, there remain some subtle issues with the sound. The best CD audio is not as relaxing or realistic as good vinyl was.
Early CD audio was even worse than it is now, by a significant margin. I could not understand why people liked it at all in 1984.
Most recordings had extremely harsh high frequency sounds, and there was a lot of aliasing. This was apparently due to errors in the early AD conversions and in rather crude analog filters that were used in the first generation players to filter out the high frequency digital artifacts which had to filter out rather high levels of noise at frequencies just above the audible range. Modern CD players have faster silicon and more transistors so they can do some sophisticated math calculations to supersample and digitally filter the audio stream and significantly lower the distortions that have to be filtered out, so they can use much less analog filtering.
It mostly works, but I still miss the "being there" feeling of the best vinyl. But I don't miss the pops and clicks. the extreme rarity of GOOD vinyl records or the fragility of vinyl. So I mostly use CDs these days.
I do agree that there is a slight edge to digitally recorded music at the 44 kHz sampling rate. I have not heard anything sampled at a higher frequency to know if this eliminates the issues. But even at 44khz, there remain some subtle issues with the sound. The best CD audio is not as relaxing or realistic as good vinyl was.
Early CD audio was even worse than it is now, by a significant margin. I could not understand why people liked it at all in 1984.
Most recordings had extremely harsh high frequency sounds, and there was a lot of aliasing. This was apparently due to errors in the early AD conversions and in rather crude analog filters that were used in the first generation players to filter out the high frequency digital artifacts which had to filter out rather high levels of noise at frequencies just above the audible range. Modern CD players have faster silicon and more transistors so they can do some sophisticated math calculations to supersample and digitally filter the audio stream and significantly lower the distortions that have to be filtered out, so they can use much less analog filtering.
It mostly works, but I still miss the "being there" feeling of the best vinyl. But I don't miss the pops and clicks. the extreme rarity of GOOD vinyl records or the fragility of vinyl. So I mostly use CDs these days.
In 1984 the average person had a music centre with a deck that made a Garrard SP25 sound positively audiophile, even pre-recorded cassettes were superior; so the sound of a CD was a revelation based on the great ITs maxim of garbage in, garbage out. In isolation, with a reasonable amp and half decent speakers, they sounded better than most were used to. For records, you had to own a minimum of a Pioneer PL12D with a good Shure cartridge to compete. Then the Japanese Mega-Amp Corporations bought up the music production companies, phased out vinyl, forcing people to buy CDs, thus increasing their profits. They are doing the same now - try buying pre-recorded VHS Tapes - in 30 years time, should I live to 90, I will still be able to play my vinyl, will the equipment still exist to play CD/SACD/HDCD or will it have been superceded by Gas Plasma Bubble Memory Players which can store the entire output of Mozart on a microdot, and play through surgically implanted piezo tranducers which produce no bass, but hey, you get a lot of music for your money!
I am not quite that cynical.
Chances are the record companies are going to be out of the loop anyway if they continue their current tactics. Bandwidth is eventually going to be so cheap that music will be able to take a wide variety of new forms, and someone will be working to actually improve things as a way of succeeding.
People would pay a monthly fee to be connected to a vast, searchable easy-to-use network. You could vote on songs you liked, and sophisticated engines could help you find others that you also liked. So artists with small audiences could find them throughout the entire world could find them and actually make a living without having a big hit.
You would no longer have to own a song to be able to play it. The issue would be how you found the songs you liked. If an artist got a few cents everytime anyone played one of their songs, it wouldn't break anyone's budget.
But this is getting way off topic, so perhaps we should move this to another thread.
Chances are the record companies are going to be out of the loop anyway if they continue their current tactics. Bandwidth is eventually going to be so cheap that music will be able to take a wide variety of new forms, and someone will be working to actually improve things as a way of succeeding.
People would pay a monthly fee to be connected to a vast, searchable easy-to-use network. You could vote on songs you liked, and sophisticated engines could help you find others that you also liked. So artists with small audiences could find them throughout the entire world could find them and actually make a living without having a big hit.
You would no longer have to own a song to be able to play it. The issue would be how you found the songs you liked. If an artist got a few cents everytime anyone played one of their songs, it wouldn't break anyone's budget.
But this is getting way off topic, so perhaps we should move this to another thread.
"The problem is that we don't want to localise (translation into American: localize) the speakers but the "phantom sources" generated by stereo reproduction.
With stereo reproduction you don't want the speakers to be localised easily. The only situuation when you would want this is if you have content that is panned to the extreme left or right."
-Charles-
My theory works on the idea that a good speaker is easily localized if played by itself in mono. When you localize sound it is from the transients in the sound correct? A loudspeaker with good imaging would then need to properly reproduce the transients in the audio signal. If the loudspeaker produces good transients in stereo, where the other speaker adds to the image with its own trasient and the mix of the two, then when played in mono it shall still have good transient properties. without the other speaker to create another sonic image it is easily localized by itself. If it plays better transients, wouldnt it be localized easier and found faster. Therefore if it were localized faster, it would have better trasienst, and better imaging
With stereo reproduction you don't want the speakers to be localised easily. The only situuation when you would want this is if you have content that is panned to the extreme left or right."
-Charles-
My theory works on the idea that a good speaker is easily localized if played by itself in mono. When you localize sound it is from the transients in the sound correct? A loudspeaker with good imaging would then need to properly reproduce the transients in the audio signal. If the loudspeaker produces good transients in stereo, where the other speaker adds to the image with its own trasient and the mix of the two, then when played in mono it shall still have good transient properties. without the other speaker to create another sonic image it is easily localized by itself. If it plays better transients, wouldnt it be localized easier and found faster. Therefore if it were localized faster, it would have better trasienst, and better imaging
Further to our earlier discussion about developing metrics and relating measured perfomance to subjective, here's some info I saw posted on another forum. OF particular interest is the weighting of the various characteristics, I think it addresses the desire for 'coefficients' one poster here had. The speaker/writer is Earl Geddes commenting on a recent AES meeting:
This seems to me to correlate pretty well with Floyd Toole's research at the NRC, which I found a little info on here:
http://www.mastersonaudio.com/features/20010115.htm
http://www.axiomaudio.com/archives/NRC.html
along with an interview:
http://www.reed-electronics.com/tmworld/article/CA475937
and a white paper here going into much more detail on Dr. Toole's current thinking:
http://www.harman.com/wp/pdf/AudioScience.pdf
I also visited Mr. Geddes's and Ms. Lee's website, and found this:
http://www.gedlee.com/distortion_perception.htm
to be of interest as well. Hope everyone else is enjoying this discussion as much as I am 😀
This seems to me to correlate pretty well with Floyd Toole's research at the NRC, which I found a little info on here:
http://www.mastersonaudio.com/features/20010115.htm
http://www.axiomaudio.com/archives/NRC.html
along with an interview:
http://www.reed-electronics.com/tmworld/article/CA475937
and a white paper here going into much more detail on Dr. Toole's current thinking:
http://www.harman.com/wp/pdf/AudioScience.pdf
I also visited Mr. Geddes's and Ms. Lee's website, and found this:
http://www.gedlee.com/distortion_perception.htm
to be of interest as well. Hope everyone else is enjoying this discussion as much as I am 😀
- Status
- Not open for further replies.