What makes the LX521 so good?

[wesayso]

I don't and never claimed any group delay number to be important, did I?
The general shape of the plot tells me way more. I tried to suggest something simple.

Do not force a linear phase correction and expect great things. But I have no question if it's audible. Let's look at a ported 3 way plus subs:

God I love that room! I wish I had something like that to play with... kudo's to jim1961.
(that ridge at 24 ms is intentional. Its a Haas kicker implemented the hard way)

Pay close attention to the scale difference. We are not talking about a few ms of difference when we look at crossover induced delay. Granted the deviation only becomes visible below 100 Hz at this scale, but it's about 2ms at 100 Hz.

The DFR:

(Delay Frequency Response)

My plot is at ~34 Hz within those same 2 ms. Jim's setup is at 15 ms at about 34 Hz.

All I'm saying is this kind of difference is indeed audible. I've already explained why that wouldn't be the case above say ~1 KHz.

 

[CharlieLaub]

All I'm saying is this kind of difference is indeed audible. I've already explained why that wouldn't be the case above say ~1 KHz.

I am very skeptical the delay curve for this plot (below) is audible. Very. Skeptical.

Compare the values to this data:

Source:
Blauert, J. and Laws, P "Group Delay Distortions in Electroacoustical Systems" Journal of the Acoustical Society of America Volume 63, Number 5, pp. 1478-1483 (May 1978)

Blauert and Laws report approximately the following thresholds for audibility:

Frequency 	Threshold of Audibility
 8 kHz                        2 msec 
 4 kHz                      1.5 msec
 2 kHz                        1 msec
 1 kHz                        2 msec
 500 Hz                     3.2 msec

In your top plot, the values are well under these levels and the curve doesn't reach 3.2 msec until about 75 Hz. The data from the published paper shows that level of delay is not audible even at 500Hz. And the lower in frequency, the less sensitive you are to group delay differences (the audible threshold is increasing rapidly below 100Hz).

What you show is a delay measurement. You don't relate that to hearing. So, again, I'm very skeptical that one can hear any audible differences due to the group delay shown in the first (top) plot in your post.

 

[wesayso]

What you show is a delay measurement. You don't relate that to hearing. So, again, I'm very skeptical that one can hear any audible differences due to the group delay shown in the first (top) plot in your post.

I showed an additional delay measurement, yes. I've played with smaller deviations than the difference between these two plots. Those very small differences were indeed noticeable. I'd rather err on the save side (more delay at 30 Hz) than have it too early. It throws off your feet automatically tapping along to the music. Feeling forced.

I know what I've heard, I've said so earlier. The shape of the plot (the other one) tells me more than the delay measurement.

Remain skeptical, no problem. I'm not trying to convince anyone but myself with my tests. I merely made a suggestion for a way to try it. But it does involve a lot of work.

Lately I've been trying to limit the effects of inter aural comb filtering. That actually works pretty good to fool our brain. It's a frequency trick more than phase. But having the phase flat and the same left and right does help to play with it. More info on my thread and the Fixing the Phantom Center thread.

I'm not trying to argue. I don't know where the limit is lower in frequency. All I know is I love to have plenty of low frequency to support the midrange. And I do notice a big difference when it's in time. How far one can stretch that? I have no clue. Better have it a bit later than too early, that's very easy to tell.

I can tell you, playing a movie with a gunshot will have you jump out of the seat. It's spooky how convincing it is, at about a 88 dB average. Play it less loud and it's not nearly as convincing anymore.

 

[Jay]

Maybe I should set up a virtual blind test with sound clips from a TP vs a 4th order LR for percussion and see if people hear the difference with headphones?

No doubt they will sound different, but this is John's argument:

aside from the phase response being difference so are the polar response, driver overlap in the stop bands, driver excursion, hence distortion, etc also different. You may prefer one speaker to the other, but to attribute it to phase response differences would be an inappropriate conclusion.

Now, lets assume we can hear the difference, and we know the theoretical differences... Of course, we need to correlate what we hear with the theory...

In many cases, the only variable that makes sense to be responsible with the audible change is PHASE. I have tried to relate audible perception with phase change and it was positive...

The problem is that the theoretical difference is "too small" to be considered correlated with audible sound change. This does not happen only with phase but also with distortion etc...

For example, why should we use input filter that has -3dB at 2Hz if theoretically it is okay if the roll off is at 20Hz. High frequency issue is more phenomenal. Why a change that theoretically only affect frequency above audible band IS audible?

YES, the difference is "small" such that most people will fail in Foobar ABX (may be I'm the only one who can pass such test) but my hypothesis that I have mentioned so many times is that even if a person doesn't pass the test, it doesn't mean that he is immune to the effect. He is just not skillful in doing the test (it requires awareness to what I call "sub-conscious" thinking process of the brain).

 

[wesayso]

I tried -3 dB at 17 Hz compared to -3 dB at 35 Hz. I know which one I prefer. I was actually hoping to prefer the latter, but without testing it I wouldn't have known. Not even to prefer the one at 35 Hz, but rather have no meaningful difference. Sadly, there was an audible difference to my ears, or rather my body and ears. We do not only hear, we feel quite a bit too.

 

[xrk971]

In many cases, the only variable that makes sense to be responsible with the audible change is PHASE. I have tried to relate audible perception with phase change and it was positive...

I think in rePhase, one can tweak the phase but leave the FR unchanged. Could one make two soundclips of this using a single full range driver and record the result and present a blind test?

With this test, the distortion, stop bands, etc will all be the same.

It will sound "phasey" for lack of a better word.

 

[xrk971]

I am very skeptical the delay curve for this plot (below) is audible. Very. Skeptical.

I think what I get from Wesayo's graph is:

My plot is at ~34 Hz within those same 2 ms. Jim's setup is at 15 ms at about 34 Hz.

The group delay at 34Hz is 15ms vs 2ms, and 2ms is basically inaudible (GOOD) and 15ms certainly is audible from standpoint of off time beats (badly synchronized foot tapping).

I like to aim for less than 5ms at 50Hz GD - as I think that is a good (inaudible) result. To achieve 2ms at 34Hz takes real work.

 

[wesayso]

I think in rePhase, one can tweak the phase but leave the FR unchanged. Could one make two soundclips of this using a single full range driver and record the result and present a blind test?

With this test, the distortion, stop bands, etc will all be the same.

It will sound "phasey" for lack of a better word.

Why use a full range speaker at all. One can convolve the original track with only a phase change. Make it an ABX test. No need to put speakers into the chain at all. You could even convolve the non phase delay track with a pure dirac pulse so we'd be sure both versions went trough the same chain.

But... and it's a big but for me.... no headphones I own give me a sense of rhythm on my eyelids, let alone other parts of my body. To me all of that helps to get a "feel" for what's going on, especially at lower frequencies.

So I'm not as convinced one would hear/sense it as easily on headphones.
All of us as enthusiastic about music should have a room like jim1961 made for himself to play with! But it might be worth a shot, but I'd suggest using the original track vs one convolved with phase rotation exhibited by a 3 or 4 way. Even that won't be a real world test, as I doubt any room without help would give the listener a clear shot of hearing it like the plots of Jim's room show. I think many (have to) deal with a lot more room sound and faults than they like to admit. What he has done is amazing (at least to me).

 

[picowallspeaker]

For example, why should we use input filter that has -3dB at 2Hz if theoretically it is okay if the roll off is at 20Hz.

I tried -3 dB at 17 Hz compared to -3 dB at 35 Hz.

I guess that Jay is talking about the time constant given by HP of the cap at amp's input
( a little bit over DC to ensure stability at open loop etc etc )

🙄

 

[phase_accurate]

There are so far no valid figures regarding the audibility of crossover induced group delay distortion IMO.
It's quite cool the Mr. Blauert and Mr. Laws did some scientific testing of group-delay audibility but there are two things that make me doubt the validity of their findings in terms of audibility of crossover induced group delay distortion:

1.)
The test signal was high-pass filtered and was therefore subject to strong group-delay distortion from the beginning. And due to the lack of low frequencies, possible effects of group delay differences between low and high frequency content might be eliminated from the start.

2.)
By the use of high-Q allpass filters they have a narrow band group delay peak which is very different from the group-delay performance of ordinary crossovers which have basically a delayed broadband low-frequency part versus a less delayed broadband high-frequency part. The test allpass filters do also have low frequency delay admittedly but it is very benign compared to the peak delay.

As soon as we have a valid model for the inaudibility of crossover-induced group-delay it would be possible to build loudspeakers that are flawless in this respect and they might probably not have to be transient perfect at all. But their crossover functions might still differ from LR.

BTW: Sebastian Goosens from the German "Institut fuer Rndfunktechnik" came to the conclusion that group delay distortions osmaller than +- 200us should be below perception in the mid and high frequency ranges:

http://forum2.magnetofon.de/bildupload/goosphase.pdf


Regards

Charles

 

[Juhazi]

I think that the major misunderstanding in this dicussion is that wesayso adjusts room response with dsp, others are talking about direct sound emitting from the loudspeaker.

 

[john k...]

Why use a full range speaker at all. One can convolve the original track with only a phase change. Make it an ABX test. No need to put speakers into the chain at all. You could even convolve the non phase delay track with a pure dirac pulse so we'd be sure both versions went trough the same chain.

Exactly what I did with a number of difference speakers. And I also compared the speaker with and without the dirac processed impulse to determine if there was any difference due to the DSP processing. In this way, no properties but the phase response changed.

 

[Jay]

So who wants to create the files for ABX?

 

[GDO]

So who wants to create the files for ABX?

You are the picky enough audiophile to bother, aren't you...😀

 

[ra7]

I already started doing this using pcd. I can put up some files for abx testing next week.

 

[BYRTT]

Regarding "threshold of Audibility" studies for group delay think not we can use exact mS numbers reported from APL_TDA for wesayso's system, we can use numbers compare wesayso's system verse jim1961's system because they both APL_TDA measurements. Engine into APL_TDA seems have unique filtering and smoothing scheme so those small numbers of mS reported for wesayso's system is very low and to show what i talk about here some comparison:

Picture 1 is 17Hz high pass BW2 with ideal opamp in free TI TINA suite.
Picture 2 is what APL_TDA report feeded a 96kHz IR-wav file created in Rephase as a 17Hz high pass BW2 filter.
Picture 3 is what REW report feeded same 96kHz IR-wav file created in Rephase as a 17Hz high pass BW2 filter.
Picture 4 is REW again scaled wider to show its not so far to TI TINA model.

In picture 5 its obvious there is unique filtering and smoothing going on into APL_TDA because that is FR window reading the same 96kHz IR-wav file into APL_TDA verse REW. Author of APL_TDA suite Raimmonds Skuruls in paper talk about use of high time resolution and there call it Time Domain Analysis (TDA), band pass filters that give high resolution timing information to see frq dependent delays in high resolution and also mention TDS as also Tom Danley talked about regarding his TEF-machine measurements in past.

 

[wesayso]

I'd say that number 4 picture comes pretty close to what I get out in my room at the listening spot. It seems to be awfully close to the BW2 at 17 Hz.

I'm still convinced headphones are no comparison to having this out in a (real) room at your listening spot. Most people I've seen posting about it on here keep insisting it isn't audible. I wasn't too sure about it myself. After hearing it, out in a room the biggest difference is the power it brings to the midrange.

It keeps you alert and on your toes. Just like a "live" performance would or does. Get it too early and it throws you off quickly. It's better to error on the late side. Even though most "think" it's easy to get this, it isn't. If it were easy why would it take me months to get it close to right. (that is, if I have it right, it might be possible to get it to behave even better, how would I know)

I bet it would be easier if the room worked with you though. There's no substitute for the kind of work jim1961 did in his room.
You may think or believe you're on a similar level but he didn't just absorb all the extra energy. All that is happening out in that room is carefully planned out. I can only dream my "living room" would perform/behave anywhere close to that.

I'm fully aware that I still have a colored sound in comparison to that room. But the timing part still is a big deal for me. It isn't number one, or even second place, like I thought it would be. Lesson learned. Even though I do like Imaging a lot, number one for me is tonality. Second place would/could be imaging. So far, still nothing to be gained with time coherency. But you often read about fast bass. What could be faster than to have the wave front at 30 Hz hit you within 4 ms?

If you're curious about your own room, just take the time to record your sound at the listening position and listen to it on headphones. It will make you (painfully) aware of "the room" you're in. If that's not enough, play back the recording though your speaker (in effect doubling your room effect). It will take you time to unlearn what you just heard. You won't notice it nearly as much listening out in that room (unless you try what I suggested here). People don't give enough credit for what their brains can do for/with that signal.

This whole phase discussion is only a small part in a bigger picture. But as such it is responsible for part of that lively feeling. Where the exact boundaries are in group delay, I don't know. I've had plenty of fun just listening before I managed to get these low numbers (and still keep it sounding right). Some other parts in the frequency domain certainly were more important (to me), like getting the low end to behave between ~ 50 and 200 Hz.

Another complaint I've read a million times is after correcting the sound, things get dull. I promise you they wont, with just a few changes. For instance we actually like sound in a room. So we actually like to have some reflections. Time them between 15 and 25 ms and they will add more than they will take away from the imaging. (remember they will take part in the tonality) Get some late reverb (probably artificially) at ~ 100/150 ms and you'll even get envelopment (pretty much impossible to achieve in a small room, without cheating). One of the things I do enjoy at concerts in good rooms, that take place in bigger venues than my living room. You won't believe how far down that signal can be and still have it work to chance your perception.

Just some random thoughts. I really shouldn't be posting this, after a night out at the TT in Assen. I might be a touch too honest here 😉. My excuses in advance.

 

[Eldam]

What is maybe difficult with your speaker is they are in corners close to the side and front walls ?

 

[GDO]

What is maybe difficult with your speaker is they are in corners close to the side and front walls ?

Disagree. Bad placement is bad excuse for bad speaker design, like M2 and stuff that require JBL ******** software like this:

SpeakerPro And SpeakerAngle | JBL Professional

 

[Eldam]

I meaned : is it not difficult to measure and time setup when the close side wall gives faster reflected sounds ?

Just measure at listener sweet spot, setup the curve and FIR et voilà ? (= any good speaker enough are getting much better with DRC and best practices ?) !

I often see the members whom get active stuffs have lesser quality amps & sources ?