Scott, I am really not that concerned about phase shift.
Evidence of audibility of moderate phase shifts is shifty at best, and that is in the sensitive band where the ability to distinguish phase for localization makes sense and is promoted by evolution (or designed in for cost saving measures, who knows for sure). Similar thing as in optics, you can only localize items larger than the pick up wavelength. The phase sensitive band therefore seems preprogrammed to be for wavelengths smaller than the size of a coffee table, and just larger than the distance between your ears.
Sound waves larger than 10 meters or so trigger something else deep down in your brain, the startle response, no need for phase discrimination. The physical response is not to look into a particular direction, but to jump up in order to find out what the heck is going on. So 8th order at 15 Hz, no problem I would think.
However, in the sensitive area, 4th order seems to be the maximum you can get away with. And I want to cross over much steeper than that. Philips had a prototype speaker in the middle of the nineties that did 8th order at 2 KHz with linear phase, flat fr, motion feedback. I was able to audition it for a couple of weeks and it was better than anything I had heard before. Smallish speaker on top of everything else. With the stuff these guys did in 1994, they should have become King of the hill, but Philips had rotten marketing.
Technical advantage of steep slopes is that you can cross over lower and that the vertical interference lobes vanish. My aim is to get my speakers as close as possible to be perceived as true point sources. This might be just the last step I need to vanquish my windmill.
Evidence of audibility of moderate phase shifts is shifty at best, and that is in the sensitive band where the ability to distinguish phase for localization makes sense and is promoted by evolution (or designed in for cost saving measures, who knows for sure). Similar thing as in optics, you can only localize items larger than the pick up wavelength. The phase sensitive band therefore seems preprogrammed to be for wavelengths smaller than the size of a coffee table, and just larger than the distance between your ears.
Sound waves larger than 10 meters or so trigger something else deep down in your brain, the startle response, no need for phase discrimination. The physical response is not to look into a particular direction, but to jump up in order to find out what the heck is going on. So 8th order at 15 Hz, no problem I would think.
However, in the sensitive area, 4th order seems to be the maximum you can get away with. And I want to cross over much steeper than that. Philips had a prototype speaker in the middle of the nineties that did 8th order at 2 KHz with linear phase, flat fr, motion feedback. I was able to audition it for a couple of weeks and it was better than anything I had heard before. Smallish speaker on top of everything else. With the stuff these guys did in 1994, they should have become King of the hill, but Philips had rotten marketing.
Technical advantage of steep slopes is that you can cross over lower and that the vertical interference lobes vanish. My aim is to get my speakers as close as possible to be perceived as true point sources. This might be just the last step I need to vanquish my windmill.
vacuphile,
It seems that once you can get the impulse response across the crossover band correct you are most of the way there. All this talk about phase shift at the low end just seems again to be a dog chasing its tail. I have been around huge sound systems outdoors and even with all those distributed bass enclosures and the distance from stage to mix console and back I wouldn't say I could fault the low frequencies, that is not where the problems show up, they always seem to be much higher frequencies that attract our attention. Nothing wrong with a properly working reflex enclosure, I've never heard anyone complain the bass was slow from the port. Now I have heard that about some seal enclosures with polypro cones though, a muddy kind of sound.
It seems that once you can get the impulse response across the crossover band correct you are most of the way there. All this talk about phase shift at the low end just seems again to be a dog chasing its tail. I have been around huge sound systems outdoors and even with all those distributed bass enclosures and the distance from stage to mix console and back I wouldn't say I could fault the low frequencies, that is not where the problems show up, they always seem to be much higher frequencies that attract our attention. Nothing wrong with a properly working reflex enclosure, I've never heard anyone complain the bass was slow from the port. Now I have heard that about some seal enclosures with polypro cones though, a muddy kind of sound.
There's a really good explanation, including clear audio demos, of the difference between linear phase and minimum phase filters. It's in the context of DAW, but very clear and should clear up a lot of misunderstandings I think.
There's a great demo of the impossible pre-sound that emerges for kick drums with linear phase filters - but loads of great and clearly audible demos of the differences and sounds. Well worth 20 minutes of time, or even watching a few times IMO:
https://www.youtube.com/watch?v=efKabAQQsPQ
Enjoy !
There's a great demo of the impossible pre-sound that emerges for kick drums with linear phase filters - but loads of great and clearly audible demos of the differences and sounds. Well worth 20 minutes of time, or even watching a few times IMO:
https://www.youtube.com/watch?v=efKabAQQsPQ
Enjoy !
This is nice but 64K length FIR's are now no problem (at least with a PC). As I always say never bet on the rest of technology stopping and sitting still. Analog once invested in a start up that was going to make 640X480 displays look "better" with primitive DSP's, of course the rest of the world was investing in making hi-res displays, at a ratio of maybe $1000/$1.
Some DSP speed info if you're interested, as you like to say it only gets better. http://www.dspguide.com/ch28/6.htm
Last edited:
Thanks I will watch this, the linear phase thing is ingrained from the way this stuff is taught. Virtually every on line tutorial goes through the usual N pole low pass filters with their "horrible" group delay if they are minimum phase.
I wish someone would counter with how non-causal linear phase filters are.
There was a guy at Apple, long gone now, who insisted that a DAC have on-chip FIR filtering. It made the chip a bunch larger and more expensive, and the IC vendor knew better than to fight it. I doubt there was any audible benefit unless you were peeking.
If this stuff gets too carried away, we will start getting annoying latencies if any real-time feedback such as in interactive video games is required.
In terms of audibility on music, a few years back (pardon if I related this already elsewhere) I was working with a group on a soundbar-subwoofer system. Of course the sub had to be wireless (why really I don't know, but it had to be). A first prototype arrived and sounded terrible, but it was hard to describe how. One supposed the deficiency would be fixed with equalization, but it didn't help. It turned out the wireless link was inappropriate and had been designed for a complete wireless multichannel system---with a latency of FIFTY milliseconds
. Without similar delay compensation of the soundbar, it was hopeless. And potentially, DVD video would have unacceptable sync with dialogue---which was a problem with early HDMI anyway.
Why do they use such dreadful samples in these sorts of things? That's a kick drum?? Give us a break ... I lost all interest in watching this after that bad start ...
Also, the level of graphic violence in these games is often very high.
S.Linkwitz is another one of those highly qualified designers who uses listening to guide his audio designs. See Group-Delay (section F) comments from Linkwitz --- he also thinks it matters at low freqs.
"Group Delay at the low end does seem to matter."
from:
Frontiers
[I bought his CD on his total work... some 500 pages, equiv.]
THx-RNMarsh
Last edited:
Richard, he says:
DSP based equalization can move Tx out in time, but how far is enough needs to be established. The aural integration time window is about 50 ms, before echoes are noted
.
The step response of the 3-way looks even further removed from the ideal and it is easy to understand why people like to think this should be accompanied by audible defects
The actual drive-unit polarity doesn't matter except when it comes to waveform preservation -- and the jury is out on that subject."
More understanding of the subject matter is obviously needed.
.
I have used headphones to convince myself that the phase distortion introduced by LR4 crossovers around 2 kHz and at 100 Hz are not audible. For example, a 1.5 kHz LR4 crossover with 24 dB/oct slopes forms a second order allpass filter with Q0 = 0.7 for its pole and zero locations.
DSP based equalization can move Tx out in time, but how far is enough needs to be established. The aural integration time window is about 50 ms, before echoes are noted
.
The step response of the 3-way looks even further removed from the ideal and it is easy to understand why people like to think this should be accompanied by audible defects
The actual drive-unit polarity doesn't matter except when it comes to waveform preservation -- and the jury is out on that subject."
More understanding of the subject matter is obviously needed.
.
I have used headphones to convince myself that the phase distortion introduced by LR4 crossovers around 2 kHz and at 100 Hz are not audible. For example, a 1.5 kHz LR4 crossover with 24 dB/oct slopes forms a second order allpass filter with Q0 = 0.7 for its pole and zero locations.
He thinks the low freq GD matters. he has his own environment and he is kinda working blind -- on his own -- in his own room etc... with is higher reverb than I would want ....and it is dated material...... But the comment of his is one of many who think it matters. Quality work. So I respect thier hunches.
Now, the real data on what is audible with GD is in another field entirely.... ear-brain-perception work. You wont find it in the AES or IEEE. There is enough of it in the book I suggested and esp with the foot-notes, will get the EE/ME people up to speed. They spell out the perceptibility of use to detetct it under best conditions. And, what conditions will have greatest affect on that. Just as in many areas of audio.. there can be masking and conditions where it is obvious and when it isnt... reverb is one and freq is another for GD. This perception stuff is tricky and just wont stay put It only confuses the issue... some hear it and some dont stuff... under non controlled conditions cant be compared well. But a lot of engineers seem to think the same way about low freq GD and that is it does matter...... but mostly only at low freqs where it is excessive.
In high-end we/me tend to err on the side of conservative. If there can be conditions where it matters, you can be assured someone out there will have that condition.... and visa versa.
I would design for what is known thru studies in other fields say is our min detectable GD amount. Then you are assured that No One can hear it under any conditions out there.
That is about all I care to say about it. Just note also that it isnt a speaker or an amp or a this or a that measured in isolation. The whole recording chain and playback gear etc added up which has to be below GD detectability. Ditto other forms of distortion. The total has to be below the threshold.
but.... there is hope....'cause ..... all together now....
'It Just Keeps Getting Better'.
THx-RNMarsh
Now, the real data on what is audible with GD is in another field entirely.... ear-brain-perception work. You wont find it in the AES or IEEE. There is enough of it in the book I suggested and esp with the foot-notes, will get the EE/ME people up to speed. They spell out the perceptibility of use to detetct it under best conditions. And, what conditions will have greatest affect on that. Just as in many areas of audio.. there can be masking and conditions where it is obvious and when it isnt... reverb is one and freq is another for GD. This perception stuff is tricky and just wont stay put
It only confuses the issue... some hear it and some dont stuff... under non controlled conditions cant be compared well. But a lot of engineers seem to think the same way about low freq GD and that is it does matter...... but mostly only at low freqs where it is excessive. In high-end we/me tend to err on the side of conservative. If there can be conditions where it matters, you can be assured someone out there will have that condition.... and visa versa.
I would design for what is known thru studies in other fields say is our min detectable GD amount. Then you are assured that No One can hear it under any conditions out there.
That is about all I care to say about it. Just note also that it isnt a speaker or an amp or a this or a that measured in isolation. The whole recording chain and playback gear etc added up which has to be below GD detectability. Ditto other forms of distortion. The total has to be below the threshold.
but.... there is hope....'cause ..... all together now....
'It Just Keeps Getting Better'.
THx-RNMarsh
Last edited:
Thought I was going mad for a bit then - the kick drum sounded so bad, tried looking at a few other YouTube videos that were reviewing kick drum mics, etc; and they weren't that great either! What's going on?!!
Found a site that had straight, full resolution samples and downloaded a few - and then played with my "good" media player ... ahh - that's better !! The combination of YouTube compression and default player were making a mess of the sound envelope, it just didn't compute to my ears ...
Found a site that had straight, full resolution samples and downloaded a few - and then played with my "good" media player ... ahh - that's better !! The combination of YouTube compression and default player were making a mess of the sound envelope, it just didn't compute to my ears ...
I'm shamefully sorry my poor quality speakers don't meet your approval.
Perhaps if you find me a set of JBL 4355s that I could
buy for a good price if I like them, I'm all ears.
OR
If you have a better recommendation that I can find
here CONUS for a good price I'm all ears.
OTHERWISE
It reminds me of my little girl when she starts to whine...
THEN
I just start counting backwards from 10...9.... She regains her composure.
She's a pretty good kid.
Cheers,
- Status
- Not open for further replies.