Hi All,
Thank you for your reactions, I really like to contribute to this forum now and then, since the forum gives me many hours of fun reading and sometimes great inspiration! (I always really like it when people totally dive into something)
Some reactions on your comments;
-regarding the room with "virtually no standing waves": of course there are always standings waves but when I design a studio control room, there will always be AT LEAST 60cm of absorbent material all around (except the floor) and at least 120cm absorbent material in the back of the room, and the speakers soffit mounted. The absorbers will always be several layers of different materials to make the impedance jump to the air as small as possible without using very expensive material such as the wedges in anechoic rooms.
I design these rooms basically to be anechoic, then I add reflectors and diffusors (only for 200Hz upwards) on top of this to create ambient, but on the sweet spot the listener will always listen semi-anechoic (no reflections, even floor reflections, within the first 20ms after the direct sound, and as diffuse as I can make it after that.
I always design the reflectors in such a way that where the engineer and the clients are sitting they will have a lot of diffuse reflections when talking or making any noise, since this really helps to not have the feeling that your in an anechoic room, which can be very unpleasant (although I personally quite like it in a strange womb kind of way ).
This way the level of the standing waves will be so much lower than the direct sound (I try to target -20dB, which depends on the size of the room and the budget of the client) That they are not really relevant any more. I do realize that this is an extreme situation for the diy community, but in my world it is pretty common. (When I read about $1000/meter speaker cabling or high-end power conditioners I always sigh and think: o my, if they would just spend 5% of that money on rockwool, the results would be orders of magnitudes bigger, but that's another topic...
-regarding my testing procedure; I do blind abx testing for myself with abx software which I wrote for that. Part of my job is helping people to decide what gear to invest in, so I also have computer-controlled relay switches coupled to the software to test things like different amps ,preamps and convertors. I really know how big the pre-bias and short acoustic memory influence can be, so I feel that it should always be tested blind this way. Normally I use a test sequence of 50 fragments to get reliable data, I always tell people that if they score less then 95%, then they can't really be sure.
I will always test with musical material which I know very well, maybe this is not correct but it works for me.
-About psycho acoustical valid testing, I don't know too much about this, and to be honest I am a bit skeptical towards it. As said my girlfriend is a psychologist and when we talk about her profession, everything is tested in ways that feel "dodgy" to me. Of course I know that it isn't really dodgy and that it has a lot of scientific ground under it, but I guess I am too much a "beta" guy, who really likes to see things expressed in numbers.
I do realize that these kind of tests are sometimes necessary, but I still am skeptical about the "trained listeners" used in some studies. Often, they appear to be students or several guys working at some audio department of a large audio company, suggesting that this means that you are a trained listener. To me a trained listener is someone who listens to sound professionally, preferable mastering engineers. The fun thing aabout mastering engineers is that often they have very well trained ears, but not much technical knowledge, resulting in describing effects in a very "normal human" way while still be able to hear it. I made some phase-linearizing filters for some mastering studios with very good environments and speakers, and you can be sure that they appreciated it! The fun thing is that they describe the effect in terms like "very deep", "next level" etc.
-regarding why non-causal filter sound bad; yes I do think that it is caused by the pre-ringing (which is long at low frequencies). But to my knowledge you can't make a non-causal situation without preringing, because then it would be causal.
-One last thought; I have the luck that I live in Amsterdam, where we have the Concertgebouw with the Concertgebouw Orchestra, often regarded as the best romantic-symphonic orchastra/room in the world. I always try to go there at least 2 times a year to do a "reality check". It keeps me on edge, realizing that music reproduction is so far away from reality that there is still basically everything to win....
If you do have the opportunity, please visit live music, especially un-amplified live music, it will make you realize how great sound really is, and what transients really are. I can assure you it will make hearing linear-phase a lot easier.
Edit: I think I should call it "minimum phase", not Linear phase, by which I mean that you don't have the phase respone caused by X-overs, but you do have the phase wrap caused by sytem hpf and lpf.
edit 2: Bruno explains it very well in this article, I fully agree with him (I know him personally): http://www.grimmaudio.com/site/assets/files/1088/speakers.pdf
Kees
Thank you for your reactions, I really like to contribute to this forum now and then, since the forum gives me many hours of fun reading and sometimes great inspiration! (I always really like it when people totally dive into something)
Some reactions on your comments;
-regarding the room with "virtually no standing waves": of course there are always standings waves but when I design a studio control room, there will always be AT LEAST 60cm of absorbent material all around (except the floor) and at least 120cm absorbent material in the back of the room, and the speakers soffit mounted. The absorbers will always be several layers of different materials to make the impedance jump to the air as small as possible without using very expensive material such as the wedges in anechoic rooms.
I design these rooms basically to be anechoic, then I add reflectors and diffusors (only for 200Hz upwards) on top of this to create ambient, but on the sweet spot the listener will always listen semi-anechoic (no reflections, even floor reflections, within the first 20ms after the direct sound, and as diffuse as I can make it after that.
I always design the reflectors in such a way that where the engineer and the clients are sitting they will have a lot of diffuse reflections when talking or making any noise, since this really helps to not have the feeling that your in an anechoic room, which can be very unpleasant (although I personally quite like it in a strange womb kind of way ).
This way the level of the standing waves will be so much lower than the direct sound (I try to target -20dB, which depends on the size of the room and the budget of the client) That they are not really relevant any more. I do realize that this is an extreme situation for the diy community, but in my world it is pretty common. (When I read about $1000/meter speaker cabling or high-end power conditioners I always sigh and think: o my, if they would just spend 5% of that money on rockwool, the results would be orders of magnitudes bigger, but that's another topic...
-regarding my testing procedure; I do blind abx testing for myself with abx software which I wrote for that. Part of my job is helping people to decide what gear to invest in, so I also have computer-controlled relay switches coupled to the software to test things like different amps ,preamps and convertors. I really know how big the pre-bias and short acoustic memory influence can be, so I feel that it should always be tested blind this way. Normally I use a test sequence of 50 fragments to get reliable data, I always tell people that if they score less then 95%, then they can't really be sure.
I will always test with musical material which I know very well, maybe this is not correct but it works for me.
-About psycho acoustical valid testing, I don't know too much about this, and to be honest I am a bit skeptical towards it. As said my girlfriend is a psychologist and when we talk about her profession, everything is tested in ways that feel "dodgy" to me. Of course I know that it isn't really dodgy and that it has a lot of scientific ground under it, but I guess I am too much a "beta" guy, who really likes to see things expressed in numbers.
I do realize that these kind of tests are sometimes necessary, but I still am skeptical about the "trained listeners" used in some studies. Often, they appear to be students or several guys working at some audio department of a large audio company, suggesting that this means that you are a trained listener. To me a trained listener is someone who listens to sound professionally, preferable mastering engineers. The fun thing aabout mastering engineers is that often they have very well trained ears, but not much technical knowledge, resulting in describing effects in a very "normal human" way while still be able to hear it. I made some phase-linearizing filters for some mastering studios with very good environments and speakers, and you can be sure that they appreciated it! The fun thing is that they describe the effect in terms like "very deep", "next level" etc.
-regarding why non-causal filter sound bad; yes I do think that it is caused by the pre-ringing (which is long at low frequencies). But to my knowledge you can't make a non-causal situation without preringing, because then it would be causal.
-One last thought; I have the luck that I live in Amsterdam, where we have the Concertgebouw with the Concertgebouw Orchestra, often regarded as the best romantic-symphonic orchastra/room in the world. I always try to go there at least 2 times a year to do a "reality check". It keeps me on edge, realizing that music reproduction is so far away from reality that there is still basically everything to win....
If you do have the opportunity, please visit live music, especially un-amplified live music, it will make you realize how great sound really is, and what transients really are. I can assure you it will make hearing linear-phase a lot easier.
Edit: I think I should call it "minimum phase", not Linear phase, by which I mean that you don't have the phase respone caused by X-overs, but you do have the phase wrap caused by sytem hpf and lpf.
edit 2: Bruno explains it very well in this article, I fully agree with him (I know him personally): http://www.grimmaudio.com/site/assets/files/1088/speakers.pdf
Kees
Last edited:
This is very similar to what I saw almost every day in the fine art printing business. Visitors to the shop were puzzled by the collections of different proofs for the same painting. "Why so many copies?" We would explain that each was a little different, we were working toward the best version. They would always exclaim "I don't see any difference." And they didn't. Until we pointed out the differences - then they would see them. They would often go on to find differences we had not seen.
Once they understood what to look for, they didn't have a hard time finding it.The next time the same person comes by the shop - same thing all over again. They don't see the subtle differences. They have forgotten what to look for and how. This would make a interesting field of study.
BTW, thanks for your ideas on phase. I'll try not correcting for the system HP and listen to what happens.
So listenable is preringing not frequency response vs phase response mismatch by itself?
Which is a personal criticism I have about Toole's published speaker tests. The norm for people, in my experience, is that they don't give the slightest damn about quality of sound as long as it isn't annoying and has enough bass! Testing loudspeakers for a population that really doesn't care may be good scientific practice but it's relevance for the small population of hifi nuts (and we are VERY much a minority) doesn't seem to hold up that I can see.
___
Earl, I forgot to say earlier: Congratulations on the retirement!
It gives more time to work with things for pleasure without having to worry about the profit, a much nicer way to go about things (if the opportunity exists) isn't it?
I took that step about a year ago, and am very glad I did. (But, I was working for a large corp, much like in a Dilbert cartoon, which made retiring even more of a relief than it might be for people who were self-employed!)
I will do some comparisons, and I may have oversimplified- white rigid foams used in packaging are not what I'm talking about. Soft, pliable layers tend to be high cell-count grey foam, which IME is acoustically absorbent. All foam I've seen sold as acoustic tile foam designed for absorption would fail your "water straight through it" test- the cells are too small. It tends to behave more as a sponge when subjected to water, due to the small cells. I'll take a closer look at known acoustic absorption foam
Foam Factory vs. Auralex - Acoustical Foam Products
and the grey, soft packing stuff I'm thinking of. Lord knows I have plenty of both, but the look and feel of them are nearly identical.
Then you have to understand my skepticism since "Linear phase" and "Minimum phase" are not only completely different, but more like completely opposite.
You should view the Toole video about psychoacoustic testing as to me, and he, nothing else means anything.
Hi Bill - retirement for me is a mixed bag because my wife is 11 years my junior and still has that long more to work, so I am trapped by the circumstances (or maybe we'd move to Oregon as well!! We love it there.)
I think that Sean Olive address the norm versus expert situation by having done tests comparing them. Both groups want the same things, its just that the pros decide on this quicker.
But yes, one reason that I stopped making speakers is that "no one cares". Its all about fad and fashion and not sound quality. And "measurements"! What's the point of doing those?
It sounds like the targeting of a minimum phase response (and an on target frequency response) via the application of excess phase correcting filters and/or linear phase crossovers implemented as an FIR filter. The full range sum of the channels involved can then move closer to the theoretical ideal of a loudspeaker (given that the speaker is not flat to dc).
I think this tends to get contracted to "linear phase" in general usage.
If the foam is support with good data then open or closed cell doesn't matter. I suppose if the cells got small enough then water would not pour through them like it does with the larger pore open cell.
That is an excellent link on acoustical foam. Very interesting seeing the marketing claims versus reality.
Again, if its minimum phase that matters then just smoothing the frequency response will do that. It may not be the phase effect at all. I seem to recall that for the most part the modal region of a room is minimum phase with few exceptions.
Yes, modes are minimum phase and can be completely equalized (see my damped SBA, page 11). But this only works if the mode doesn't overlap with other modes. It must occur separated. A SBA does this, because it reduces the three dimension to only one.
In common stereo setups there is much overlapping and in this case no minimum phase behaviour anymore.
Which is very similar to what I was showing with my 2 examples earlier.
One linear phase (phase extends to DC)
and the other one minimum phase for the FR I chose to correct.
In both cases I use filters that are partly linear phase to get this correction (at the listening position) and they have very similar pré-ringing (but it is down by quite a lot of dB's).
Yet the correction following minimum phase for the total frequency response I cover (second graph) sounds most natural to me. So I was glad to read I'm not alone in this preference. Indeed most refer to this as linear phase as it has no phase rotation from crossovers in the response. My solution does not use crossovers but still needed help to get it this straight at the listening position.
Sorry no, if you insert an allpass filter in a chain you can have a perfectly flat frequency response with loads of phase shift.
By minimum phase I mean (and probably my definition term is wrong here) that all x-over points are linear phase, and the resulting phase curve is the minimum-phase curve which corresponds to the (is causal to the) total system frequency response.
So this would be the phase curve if you used a Hilbert transform with the frequency response as input. You can do this in REW (Wezajo showed this in hist post)
...This is what I meant by the quality of my English
(or maybe my math skills)Please read the link to the Bruno Putzeys article I posted, this is excatly what I mean, and he probably explains it better.
Kees
Which is very similar to what I was showing with my 2 examples earlier.
One linear phase (phase extends to DC)
and the other one minimum phase for the FR I chose to correct.
In both cases I use filters that are partly linear phase to get this correction (at the listening position) and they have very similar pré-ringing (but it is down by quite a lot of dB's).
Yet the correction following minimum phase for the total frequency response I cover (second graph) sounds most natural to me. So I was glad to read I'm not alone in this preference. Indeed most refer to this as linear phase as it has no phase rotation from crossovers in the response. My solution does not use crossovers but still needed help to get it this straight at the listening position.
Hi Wesayo, we where posting at the same time it seems.
Yes you understand what I mean, but it looks in your second plot that you made the phase response 12dB/oct instead of 24 dB/oct which the frequency response does.
If you try this I think it will sound even better.
If executed correcly, your acoustical measured response should have no preringing.
Kees
Last edited:
have you compared to a setup that has the same FR but without the (excess) phase correction?
personally I use a similar (excess phase) correction but haven't done that comparison rigorously so hard for me to say whether the the phase correction has a marked impact or not
This doesn't make any sense to me. If two modes are minimum phase then why if they overlap would it be non-minimum phase? What makes them non-minimum phase?
Since, in theory, all modes overlap to some extent, when do they suddenly become non-minimum phase.
Sorry, but this just doesn't sound right.
The overall result is not minimum phase. It is the same as two ways crossed at a certain frequency. Each way is minimum phase. Together they are not. It is not possible to equalize both amplitude and phase anymore with a minimum phase filter.
But then its not minimum phase any more. I am confused and maybe its a language problem, but something here doesn't sound right.
Also in you blind subjective tests how did you setup the test for detecting phase? Using headphones or speakers?
Hello Kees,
The grey phase trace is the generated phase as calculated by REW from the FR plot. The red trace is the measured phase. I have a downward sloping FR which was covered in another thread (before anyone goes there
). I do not "draw" the phase curve but get there trough changing parameters in DRC. So it will be harder to try something different compared to people using RePhase for instance.It seems my goals in my livingroom have a lot in common to your preferred listening environment. The clean first ~20 ms, late diffused reflections after that. I do use some other tricks or features but that's another thread entirely.
Yes I have. it did not start out like these plots for me. Got a minute (or several hours lol)? Most of it is covered in my own thread: http://www.diyaudio.com/forums/full-range/242171-making-two-towers-25-driver-full-range-line-array.html
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.