To EQ or not to EQ
The argument I have often heard advanced against deliberate equalization is that it introduces errors in the phase response, or the time response.
What is not understood, is that when deviations from uniform (aka flat) response occur, they have already introduced errors in the phase response and the time response, and that using EQ to correct these deviations actually restores all three at the same time: the frequency response, the phase response, and the time response (and this is because they are all, essentially, aspects of the same thing...the signal itself).
Now, this is generally true, which means in some cases it may not be true. But in specifically what many people believe is the most troublesome case, room EQ (to uniform response), famous TAS audio reviewer (and Professor of Mathematics) Robert Green claims that room response deviations from linear reflections, which most are, are minimum phase, and hence are corrected by the inverse deviation.
So far from being horrible, correct EQ is beneficial to the phase and time.
Now this is not to exaggerate the importance of phase response. Large volumes of research show it is not audible to surprisingly large rotations. Anyway. In all cases like this, existing research doesn't prove it won't be audible at this or that time. But, if you want to prove you can consistently hear it, it's probably going to be hard enough that you wouldn't bother. So, if it so hard to hear, it can't be too horrible, can it? Not to say the challenge of getting perfect time response, hopefully w/o sacrificing too much else, and appreciating it as one can, can't be a fun endeavor on it's own. But existing knowledge doesn't suggest it's worth sacrificing many other things, notably the frequency response, for. In most cases.
Then there is the matter of what you might call, deliberate non-uniform response. Now, those indeed will be adding time and polarity deviations, along with the frequency response variations. But this is the natural way these things occur, and are generally accounted for in only describing the frequency response (not the "horror" of deliberate EQ in particular). So if you were, say, achieving the same result with some tweaky infidelity, exactly the same and not added distortion or whatever, your tweaky response correct would have the exact same effect on the phase and time as a deliberate EQ system--preferable a digital EQ but some high end analog EQ's are truly high fidelity also (such as the Cello Pallete).
Now, before the days of digital EQ, analog EQ's had lots of practical issues, including but not limited to distortion causing devices. It was rare to find a really good one; before the Palette, I'm not sure there was. Now, with digital input and digital output, a cheap digital one can be as perfect as the 24/96 audio it puts out (in the case of the Behringer 2496) when you use full digital I/O. The distortion is many zeros below the decimal point, the noise level is better than -144dB. Jitter does not appear to be any higher in the output than in the input...it does not appear to add to jitter (nor does my whole chain of processors and converters)--and once again not to exaggerate the importance of jitter, but just saying. And of course, precisely the filter you wanted, with no pole out of place.
Designers of loudspeakers have long understood this, and the importance of compensating for driver impedance and response in the crossover is done. This is just another form of deliberate EQ, often with somewhat precision-limited parts, but often very carefully considered designs ("engineered"). And correcting the driver response corrects the phase as well. However, most crossovers do introduce group delay and some time dispersion. Virtually all scientists, engineers, and manufacturers do apply more crossover than the minimum: just lowpass on the tweeter, the 1930's standard.
The primary issue involved here, IME, is loudness uncompensated listening comparisons. Without compensating for the difference in efficiency, a more minimal crossover will play louder, and hence have a large tendency to sound better. If drive power can be increased without issue, the less efficient corrected response, done correctly, would be preferred. Not that there aren't limits as to how far this can be done without diminishing dynamic range.
Since the scientists, engineers, and manufacturers generally do level adjusted comparisons (though perhaps within a desired range of efficiency) they find the more complex crossover to sound better.
Subjectivist gurus do not do level matched comparisons, so the more minimal crossover sounds better.
The same issue, typically combined with sighted expectation bias and other biases, explains a lot of tweako beliefs, from power conditioners (maybe not a bad idea, but not with the massive benefits claimed), expensive audiophile cables (which if they do anything at all to the sound is probably destructive to simple well made cables), lack-of-feedback, single ended, and NOS.
In every case, if there are actual changes, people may find they prefer them, for highlighting this or that, but perhaps also what people find they prefer in casual testing might not over the long term be appreciated as well.
I think it's very risky to try to achieve fidelity through infidelty, other than correctly determined and deliberately applied EQ and possibly out-of-band filtering.
But, it's your roll.
Now, why does loudness, even trifling amounts, make things sound better?
Because it increases our auditory S/N, and opens up a far greater amount of information.
Now one thing to be clear is we never fully absorb the information of even the most limited presentation. So why do we need more information?
Because it gives our auditory system a greater sample to choose from; it opens up patterns that previously were hidden by noise (most often...household ambient noise).
Now, there are limits to this that vary a lot, and playing so loud as to cause compression (not to mention clipping) may be undesirable (especially, compression in your own hearing system). And as I posted before, actually avoiding current or voltage limiting may be far harder than people recognize.
Which is another thing. Unless you have very high efficiency speakers, or even so, you may need far more power than you realize. In practice, little spikes of peak power are required up to alarming amounts.
Sanders' estimate was 500W for typical dynamic loudspeakers.
That may be a bit high for typical dynamic speakers. I would think 100W for kiddie speakers, 200W for serious speakers, and 300W for inefficient high end speakers.
But there I am, giving advice again.
I see John Curl describes his friend Dick Sequerra's Metronome speakers as "kiddie speakers." They're small, with small drivers, though possible do require more than 200W amplifier.
Anyway, they take clean impulse response to an exteme. The crazy looking tweeter and weird woofer do actually deliver very clean impulse response, hence the name Metronome. However, the frequency response is way out of whack, and finding the right "location" may be problematic too.
This is the example of sacrificing much to get the perfect impulse response, an ultimate tweako goal (since the serious scientists, engineers, etc, don't consider it THAT important, though never discounted entirely, and there is some room for doubt).
They do have amazingly well focused imaging, though not all people like that either. But many people who try them for awhile find they dislike some aspect of this approach. I'm sure I couldn't stand it at all, though it might be interesting to have a pair for an imaginary magic show.
So that's what they are. Stage magic equipment. A good description also of high end tweako stuff which is less well engineered.
The argument I have often heard advanced against deliberate equalization is that it introduces errors in the phase response, or the time response.
What is not understood, is that when deviations from uniform (aka flat) response occur, they have already introduced errors in the phase response and the time response, and that using EQ to correct these deviations actually restores all three at the same time: the frequency response, the phase response, and the time response (and this is because they are all, essentially, aspects of the same thing...the signal itself).
Now, this is generally true, which means in some cases it may not be true. But in specifically what many people believe is the most troublesome case, room EQ (to uniform response), famous TAS audio reviewer (and Professor of Mathematics) Robert Green claims that room response deviations from linear reflections, which most are, are minimum phase, and hence are corrected by the inverse deviation.
So far from being horrible, correct EQ is beneficial to the phase and time.
Now this is not to exaggerate the importance of phase response. Large volumes of research show it is not audible to surprisingly large rotations. Anyway. In all cases like this, existing research doesn't prove it won't be audible at this or that time. But, if you want to prove you can consistently hear it, it's probably going to be hard enough that you wouldn't bother. So, if it so hard to hear, it can't be too horrible, can it? Not to say the challenge of getting perfect time response, hopefully w/o sacrificing too much else, and appreciating it as one can, can't be a fun endeavor on it's own. But existing knowledge doesn't suggest it's worth sacrificing many other things, notably the frequency response, for. In most cases.
Then there is the matter of what you might call, deliberate non-uniform response. Now, those indeed will be adding time and polarity deviations, along with the frequency response variations. But this is the natural way these things occur, and are generally accounted for in only describing the frequency response (not the "horror" of deliberate EQ in particular). So if you were, say, achieving the same result with some tweaky infidelity, exactly the same and not added distortion or whatever, your tweaky response correct would have the exact same effect on the phase and time as a deliberate EQ system--preferable a digital EQ but some high end analog EQ's are truly high fidelity also (such as the Cello Pallete).
Now, before the days of digital EQ, analog EQ's had lots of practical issues, including but not limited to distortion causing devices. It was rare to find a really good one; before the Palette, I'm not sure there was. Now, with digital input and digital output, a cheap digital one can be as perfect as the 24/96 audio it puts out (in the case of the Behringer 2496) when you use full digital I/O. The distortion is many zeros below the decimal point, the noise level is better than -144dB. Jitter does not appear to be any higher in the output than in the input...it does not appear to add to jitter (nor does my whole chain of processors and converters)--and once again not to exaggerate the importance of jitter, but just saying. And of course, precisely the filter you wanted, with no pole out of place.
Designers of loudspeakers have long understood this, and the importance of compensating for driver impedance and response in the crossover is done. This is just another form of deliberate EQ, often with somewhat precision-limited parts, but often very carefully considered designs ("engineered"). And correcting the driver response corrects the phase as well. However, most crossovers do introduce group delay and some time dispersion. Virtually all scientists, engineers, and manufacturers do apply more crossover than the minimum: just lowpass on the tweeter, the 1930's standard.
The primary issue involved here, IME, is loudness uncompensated listening comparisons. Without compensating for the difference in efficiency, a more minimal crossover will play louder, and hence have a large tendency to sound better. If drive power can be increased without issue, the less efficient corrected response, done correctly, would be preferred. Not that there aren't limits as to how far this can be done without diminishing dynamic range.
Since the scientists, engineers, and manufacturers generally do level adjusted comparisons (though perhaps within a desired range of efficiency) they find the more complex crossover to sound better.
Subjectivist gurus do not do level matched comparisons, so the more minimal crossover sounds better.
The same issue, typically combined with sighted expectation bias and other biases, explains a lot of tweako beliefs, from power conditioners (maybe not a bad idea, but not with the massive benefits claimed), expensive audiophile cables (which if they do anything at all to the sound is probably destructive to simple well made cables), lack-of-feedback, single ended, and NOS.
In every case, if there are actual changes, people may find they prefer them, for highlighting this or that, but perhaps also what people find they prefer in casual testing might not over the long term be appreciated as well.
I think it's very risky to try to achieve fidelity through infidelty, other than correctly determined and deliberately applied EQ and possibly out-of-band filtering.
But, it's your roll.
Now, why does loudness, even trifling amounts, make things sound better?
Because it increases our auditory S/N, and opens up a far greater amount of information.
Now one thing to be clear is we never fully absorb the information of even the most limited presentation. So why do we need more information?
Because it gives our auditory system a greater sample to choose from; it opens up patterns that previously were hidden by noise (most often...household ambient noise).
Now, there are limits to this that vary a lot, and playing so loud as to cause compression (not to mention clipping) may be undesirable (especially, compression in your own hearing system). And as I posted before, actually avoiding current or voltage limiting may be far harder than people recognize.
Which is another thing. Unless you have very high efficiency speakers, or even so, you may need far more power than you realize. In practice, little spikes of peak power are required up to alarming amounts.
Sanders' estimate was 500W for typical dynamic loudspeakers.
That may be a bit high for typical dynamic speakers. I would think 100W for kiddie speakers, 200W for serious speakers, and 300W for inefficient high end speakers.
But there I am, giving advice again.
I see John Curl describes his friend Dick Sequerra's Metronome speakers as "kiddie speakers." They're small, with small drivers, though possible do require more than 200W amplifier.
Anyway, they take clean impulse response to an exteme. The crazy looking tweeter and weird woofer do actually deliver very clean impulse response, hence the name Metronome. However, the frequency response is way out of whack, and finding the right "location" may be problematic too.
This is the example of sacrificing much to get the perfect impulse response, an ultimate tweako goal (since the serious scientists, engineers, etc, don't consider it THAT important, though never discounted entirely, and there is some room for doubt).
They do have amazingly well focused imaging, though not all people like that either. But many people who try them for awhile find they dislike some aspect of this approach. I'm sure I couldn't stand it at all, though it might be interesting to have a pair for an imaginary magic show.
So that's what they are. Stage magic equipment. A good description also of high end tweako stuff which is less well engineered.
No comments:
Post a Comment