Geesh! Guys, the point isn't the simulations. 🙂 I'm just wondering if time alignment really matters, and I used those step plots to explain what I meant between phase aligned, and phase and time aligned speakers since it seems I was not making myself clear.
Maybe I should have said speakers with no group delay between drivers. << sigh >> 🙂
As I'm sure you know, some vendors have made a big deal about this distinction and it's benefits. I wanted an unbiased view of whether it was snake oil or not.
Best,
Erik
Maybe I should have said speakers with no group delay between drivers. << sigh >> 🙂
As I'm sure you know, some vendors have made a big deal about this distinction and it's benefits. I wanted an unbiased view of whether it was snake oil or not.
Best,
Erik
Last edited:
So here are some charts from a Stereophile article written by John Atkinson on measuring speakers, what little of it they do.
First is what JA calls "time-coherent: example plot. Thank god, he doesn't explain the crossover topology, or specify the exact speaker it comes from:
Immediately afterwards he shows the step response for a more conventional design.
Irrelevant of the crossover topology or frequency response, I'm just asking if anyone has experienced that being "time-coherent" made much of a difference. My only experience with a time-coherent speaker was unfortunately with speakers with such terrible treble response I could not stand to be in the same room with even if off, for fear of the damage they'd do to my hearing. 🙂 So I have no good frame of reference that says it's worthwhile trying.
Do you? 🙂
Thanks!
Erik
P.S. Oh, wait, I also recently heard the Mundorf Kit speakers, but damn, the room, electronics and music were unbelievably bad. I attribute nothing I heard to those speakers.
First is what JA calls "time-coherent: example plot. Thank god, he doesn't explain the crossover topology, or specify the exact speaker it comes from:
An externally hosted image should be here but it was not working when we last tested it.
Immediately afterwards he shows the step response for a more conventional design.
An externally hosted image should be here but it was not working when we last tested it.
Irrelevant of the crossover topology or frequency response, I'm just asking if anyone has experienced that being "time-coherent" made much of a difference. My only experience with a time-coherent speaker was unfortunately with speakers with such terrible treble response I could not stand to be in the same room with even if off, for fear of the damage they'd do to my hearing. 🙂 So I have no good frame of reference that says it's worthwhile trying.
Do you? 🙂
Thanks!
Erik
P.S. Oh, wait, I also recently heard the Mundorf Kit speakers, but damn, the room, electronics and music were unbelievably bad. I attribute nothing I heard to those speakers.
Last edited:
I think if you want results of a careful test like that you won't find it. For what it's worth (almost nothing since it's just an uncontrolled loosely done test with a sample size of one), I have Synergy-based speakers that were with a lot of effort designed to be linear phase.
Linear phase equates I think to what you are calling time aligned. Linear phase does a visually recognizable job of reproducing time domain waveforms like square waves over some region of the audio range. To do that, the magnitude response has to be somewhat flat and the phase response, when excess delay is removed, has to stay within some narrow range of degrees (perhaps 30) near zero.
I first designed the speakers with just a crossover to get them playing. Then I put in stupid amount of time (even developed my own crossover simulator program so I could work with crazy structures like allpass filters) and got them to have more or less linear phase. The time between hearing these, though, was hours, but the difference wasn't exactly like a flock of angels descended on the listening room radiating audio amazingness. At best it was "yeah, I think that might sound better, maybe". And that with crossover slopes changed and magnitude flatness a little different so meaningfulness of this report goes even further down the drain!
I like the idea intellectually, though (seems the pressure waveforms should be true to original doesn't it?), so I went further and added a FIR type DSP crossover to flatten the magnitude and phase responses even further. But, again no revelation at least in the high freqeuencies -- sounds a little smoother, but then the dB curve is flatter, so you'd probably expect that. The FIR DSP did let me EQ down into lower frequencies (which with room and baffle step effects get really crazy in the phase response), and that does seem to be audible, particularly with some (though not too many) percussion strikes. But still, no controlled test, and only me as the subject, so....
And just because I like to show them off, here are unsmoothed frequency and impulse response of the speakers (before FIR)
And a square wave sweep can be found at
https://www.youtube.com/watch?v=YVOM7GMkg28
If you want to play the sweep through your own speakers and look though a microphone and oscilloscope (or equivalent), here is the file (big download):
30 seconds of squarewave sweep WAV file
Linear phase equates I think to what you are calling time aligned. Linear phase does a visually recognizable job of reproducing time domain waveforms like square waves over some region of the audio range. To do that, the magnitude response has to be somewhat flat and the phase response, when excess delay is removed, has to stay within some narrow range of degrees (perhaps 30) near zero.
I first designed the speakers with just a crossover to get them playing. Then I put in stupid amount of time (even developed my own crossover simulator program so I could work with crazy structures like allpass filters) and got them to have more or less linear phase. The time between hearing these, though, was hours, but the difference wasn't exactly like a flock of angels descended on the listening room radiating audio amazingness. At best it was "yeah, I think that might sound better, maybe". And that with crossover slopes changed and magnitude flatness a little different so meaningfulness of this report goes even further down the drain!
I like the idea intellectually, though (seems the pressure waveforms should be true to original doesn't it?), so I went further and added a FIR type DSP crossover to flatten the magnitude and phase responses even further. But, again no revelation at least in the high freqeuencies -- sounds a little smoother, but then the dB curve is flatter, so you'd probably expect that. The FIR DSP did let me EQ down into lower frequencies (which with room and baffle step effects get really crazy in the phase response), and that does seem to be audible, particularly with some (though not too many) percussion strikes. But still, no controlled test, and only me as the subject, so....
And just because I like to show them off, here are unsmoothed frequency and impulse response of the speakers (before FIR)
https://www.youtube.com/watch?v=YVOM7GMkg28
If you want to play the sweep through your own speakers and look though a microphone and oscilloscope (or equivalent), here is the file (big download):
30 seconds of squarewave sweep WAV file
Yup.
I use 24db LR active crossover and slide the the tweeter till the centers are lined up.
It just sounds better.
@2khz I had a hard time aligning the phase, but you could hear it much easier once the tweet was slid back so the centers were close.
I think 6db crossovers are best if you listen at lower volumes.
I like full range drivers and older thiel speakers, till I need to crank it.
I use 24db LR active crossover and slide the the tweeter till the centers are lined up.
It just sounds better.
@2khz I had a hard time aligning the phase, but you could hear it much easier once the tweet was slid back so the centers were close.
I think 6db crossovers are best if you listen at lower volumes.
I like full range drivers and older thiel speakers, till I need to crank it.
Did you look at the plots I linked to? I wasn't only talking about crossover topology you know 🙂.
It took me almost a year to find my own specific recipe (that still may change). Do not look at plots alone but listen too. I can hear the difference between linear phase and minimum phase corrected speakers, FR being the same between the two. I did more than just casual tests of this phase behaviour. To me this definitely isn't snake oil. Is it a large difference? To me it is, but not as large as a flat FR. But you've got to work on more than the speakers alone. Your room has to support the efforts. If you browse trough my thread you can see and follow the journey I made.
Think of the speakers and room as one system.
I use frequency dependant windows for the correction, it did matter a lot what length was chosen at which frequency. FR being the same between those plots. I tried anything I could come up with and finally narrowed down to the FDW I liked best. I can tell you that minor differences here can make a big change. But you'll hear it when you get it right. I don't ask for anyone to blindly trust that though. Frankly I don't care. Its the subject of discussion here so I shared my experience.
The plot you post here from Stereophile? That was measured at (too) close distance and isn't a representative of what that particular speaker is capable off. Yes I do know which speaker that is. That's how crazy I was to find my own answers.
In all honesty, I'm playing with ambient speakers (have been for a while) to change the way my room sounds, you have to be pretty deaf to miss that change! I'm not trying to add anything (even though technically I am) but it's more like hiding my own room and substitute it for a better one. To do that you'd have to have a reasonably controlled room result to start with.
I like the combination of both but as said, this ambient project is a way bigger or more obvious change. Phase linearization is more subtle compared to that.
I like both. 🙂
my experience with multi-way pa systems is slightly different being that in some cases the bass,mid and high frequency systems where all horn types that have internal path length differences so the ability to time align the sections does have a significant impact on overall response and can affect the coverge pattern (lobing at x-over) for me transients like percussion have far more impact with a time aligned system (which to my understanding goes a long way in correcting phase issues) for home speakers with physically smaller distances i do think that it still applies otherwise the time alignment of studio monitors would be an exercise in futility.
Yes, I can turn it on and off with a click in Jriver.
I have a standard LR36 crossover @ 800hz that is in phase in the XO region but has the usual phase wrap / group delay versus rest of the band.
Then I have a phase correction filter made in rephase that is convoluted into the signal that can be turned on and off almost instantly.
To my big suprise the difference is pretty easy to notice in favour of the one with flat phase, though it depends a little bit on the type of material.
Percussion instruments sounded more real, the speaker sounded more seamless (like one driver) and vocal ineligibility increased .
Test was done in mono, just one speaker. So don't really know how it will hold up in stereo, but I suspect it will and that it will also affect imaging.
The frequency respons is identical i both cases. The impulse response looks better in the version with corrected phase.
Hi Fredrick,
There is nothing special about stereo when it comes to that kind of testing. Mono makes it easier to notice the differences because you loose the effect that can easilly distract you. If it sounded right in mono, it will sound right in stereo.
All of the tests done by JBL is done in mono and with one loudspeaker in front of the listener so I think your methods are spot on.
The last two posts pretty much coincide exactly with my experience.
Having zero time misalignment in the all driver crossover regions as well as the same vertical and horizontal coverage angles is extremely audible--it's not close.
If you're using direct radiators, the problem of course is early reflections from around the loudspeakers disguising direct sound arrivals. Your presented sound stage image will be confused enough to not notice the differences. Most people move their loudspeakers away from the walls to compensate for this, but this also significantly adds to bass modulation distortion after boosting the LF to compensate for the loss of bass. This is also audible as added "muddiness", especially at SPLs above ~80 dBC.
Chris
Having zero time misalignment in the all driver crossover regions as well as the same vertical and horizontal coverage angles is extremely audible--it's not close.
If you're using direct radiators, the problem of course is early reflections from around the loudspeakers disguising direct sound arrivals. Your presented sound stage image will be confused enough to not notice the differences. Most people move their loudspeakers away from the walls to compensate for this, but this also significantly adds to bass modulation distortion after boosting the LF to compensate for the loss of bass. This is also audible as added "muddiness", especially at SPLs above ~80 dBC.
Chris
Hi Bill
I concur with what you are saying.
I used to not think that phase linearity was a big deal at all. Many test (possible poorly done) did not show anything of significance. To wit the best being those of Brian Moore whose tests are incontrovertible, but whose results are somewhat limited to very specific unique situations. Hard to extrapolate to the more general case, but within their range of applicability showed no importance to group delay.
But then I was reading some of Griesinger's works and it all clicked to me. He claims that linear phase is important from about 700 Hz to 7000 Hz, precisely where most crossover do not have linear phase. I started thinking and doing some testing.
I found that my own speakers, despite doing this intentionally, where very nearly linear phase. They have flat listening axis response, flat power response (also known as a flat DI) and linear phase. They are the best speakers that I have ever heard. So maybe linear phase is part of the equation. Not much real evidence to support that either way however.
To me time and phase are the same things for every time there is a unique phase for any given frequency. Time is not frequency dependent and so it is easy to talk about. Phase is frequency dependent so it is more complex situation, but sometime more pertinent.
Linear phase across the entire frequency range does not seem to be necessary.
Well designed test. Agrees with what I was saying - it is a factor, but not a dominate one. Frequency and polar response seem to be more important, but once you get those under control then phase might be important to optimize the system.
Erik,
I certainly believe that some can hear the difference between a speaker with flat phase compared to wrapped phase, but I don't seem to be able to.
Assuming you are unbiased, why not listen to the video I linked in post #13, and hear the difference for yourself?
Best to use a good set of headphones for the comparison.
Art
I might try that, though I loathe JRiver's User Interface. All I have is OmniMic though. How easy is it to go from an OmniMic measurement to a Rephase equalized setup?
Best,
Erik
Best,
Erik
Some special character attracts people to FR drivers… perhaps the built-in time coherence?
dave
A lot more difficult than listening to the comparison in post #13 😉.
I purchased a BSS Blu 100 so I could implement FIR filters, but after listening to the comparison, now have little desire to go through the process knowing I wouldn't be able to hear the difference...
Totally different sound through my stax headphones, sounds very much like going from 1st order in multi-horns to 4th order - it's annoying to me .. lol - neat recording thx
Yes, frequency response and polar response is way more dominant. But it surprised med I found it as audible as I did. I have read many times that such small group delays should be inaudible.
The difference was subjectively bigger than many other things I and many other DIYers usually worry about.
To exclude bias I need to do this blind tough. I probably will at some point.
I also don't really know if there could be other technical reasons the convoluted signal sounds different? As mentioned frequency response/ SPl was identical conirmed with measurement...
We showed that group delay audibility increased with SPL, Moore also agrees with this. Very few tests control SPL making the results very hard to confirm.
Also few loudspeakers are going to have a good enough FR and polar response to allow for group delay detection.
Will offer a bit of help, measure with your Dayton mic into REW, zero align IR with command "Estimate IR delay", then filter with FDW 1/6 octave (4,3mS) and upload a exported IR-wav file here. Will then massage it in Rephase and upload a IR-wav file ready for loading into JRiver convolution container.
Below exercise speaker 10F8424/SPH-250KE FAST system is aligned nearfield but even it looks crazy when going farfield alignment means a lot for perceived sound and realism. Correction filter into JRiver convolution container correct for high order LR8 at 355Hz and in PEQ container tweeter is delayed 0,4mS.
Attachments
Last edited:
- Status
- Not open for further replies.