• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

What is your favorite house curve

mitchco

Active Member
Joined
May 24, 2016
Messages
105
Likes
185
#62
@Chipless here is the corrected target responses:

frequency response targets fixed.JPG


While it looks like there is more variability in targets, that is not how it comes out in the measurements. Here is a comparison of using the OliveToole target (i.e. +2 dB @ 20 Hz with a straight line to -8 dB @ 20 kHz) in Audiolense versus the Flat to 1 kHz and straight line to - 6 db @ 20 kHz target in Acourate.

Audiolense and Acourate comparo - different targets.jpg


The green and red trace is Audiolense and the blue and purple is Acourate, using the two different targets as specified above. Note these are two different speaker systems in the same room measured years apart. The red and green trace is a two way floorstander with dual subs and the blue and purple trace is a 3 way floorstander. All active and large speakers.

The two systems measure closer together than the target definitions specify. In fact, above Schroeder, they are virtually identical, even though the target's used say different. Why? There is some variability in the two DSP software packages used. Meaning each package uses a slightly different analysis technique and psycho-acoustic filtering. The details of which are discussed in my book for Acourate. For Audiolense, one can read about it in the help file included at the bottom of this article.

For me, what matters is the tonal response or timbre to my ears. Even with different speakers using different DSP packages over a period of a couple years, I consistently measure the same tonal preference...
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
2,162
Likes
864
Location
UK
#63
This thread's premise is in serious conflict with what is being said in the Genelec thread...
 

mitchco

Active Member
Joined
May 24, 2016
Messages
105
Likes
185
#64
@Cosmik I don't think so. Actually, I am in agreement with @Floyd Toole Let's see if I can summarise and show where there might be a difference or two.

- start with a loudspeaker that measures well in an anechoic environment - I am certainly grateful for the spins and really wish many other speaker manufacturers would take this up. This is the crux of the problem. The simple reality is that the (vast) majority of speakers are not well designed.

- it helps if the off axis response is similar to the on-axis response. I use constant directivity waveguides. I think Dr. Toole, Dr. Olive at Harman have done a wonderful job with the spinorama system and subjective listening tests, with the ability to predict a good sounding loudspeaker based on the anechoic data.

- below Schroeder the room comes into play and could use some assistance. Harman and several other vendors offer hardware and software solutions to assist. Some do it better than others.

- time alignment/time domain correction - I wish there was more research performed in this area. In my limited subjective listening experiments in-room, I can create a filter that has identical frequency response, but one filter is time domain corrected and the other not. In a controlled experiment, I can hear a subtle difference and reference that in my latest article on CA. I am humble enough to admit that I could be "hearing things" and hence would love to see more research. But my results seem to correlate with others that have performed similar experiments. Also note all of my experiments are using linear phase digital XO's. There are no passive XO's in the experiments I performed.

- eq above Shroeder. This is a possible area of divergence. As Dr. Toole points out, a well designed loudspeaker should not require eq above Schroeder. If I had Salon2's or JBL M2's in my living room, it is quite likely that I would not use any eq, based on the spins I have seen. But this is the crux of the issue Cosmik. Really well designed loudspeakers like the Revel's and JBL's and likely the D&D 8C's are far and few between. The vast majority are not well designed. Some have horrible off-axis response (i.e. uneven directivity) and for sure eq can't fix that. That's why I use constant directivity waveguides. Poorly designed passive XO's that produce lobing, again can't be fixed with eq. Hence why I use linear phase digital XO's.

But here is the biggy - the vast majority of speakers to my ears are (way) too bright sounding and directly correlates to the in-room steady state measurements I have taken. One typical example is the Dynaudio 600 XD's If you look you can see the in-room measurement compared to my reference is some 5 dB hotter from about 3 kHz on up. Way too bright for my ears. In fact, I can't even listen to it. So what to do? Unfortunately, shelving tone controls don't work as what one needs is a tilting tone control. And if there were such a thing, I could probably be happy. But since I am already using DSP to assist below Schroeder, I might as well use it for a tilting tone control. Which is what it is doing above Schroeder as it is mostly the direct sound that is being eq'd. In combination with linear phase digital XO's and constant directivity waveguides, the off axis response is almost identical as the on-axis response. I show that in detail my book.

Cosmik, you can pick this apart, but I hope you can see that while the approach is different, the end result is similar. I can't afford Salon 2's or M2's at the moment, but I can afford some sophisticated DSP that assists in smoothing out the bottom end and eq's the tonal response (i.e. direct sound timbre) to something that I prefer without having my ears ripped off from too much high frequency energy. As mentioned, I am hopeful that other speaker manufacturers start leveraging the great research from Dr. Toole and Dr. Olive and start producing better sounding speakers.
 

Sal1950

Major Contributor
The Chicago Crusher
Joined
Mar 1, 2016
Messages
4,125
Likes
1,235
Location
Central Fl
#65
So if I decide to set Audyssey editor to not eq above X freq, where would you guys recommend I set it then, around 200 khz?
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
2,162
Likes
864
Location
UK
#66
@Cosmik I don't think so. Actually, I am in agreement with @Floyd Toole Let's see if I can summarise and show where there might be a difference or two.

- start with a loudspeaker that measures well in an anechoic environment - I am certainly grateful for the spins and really wish many other speaker manufacturers would take this up. This is the crux of the problem. The simple reality is that the (vast) majority of speakers are not well designed.

- it helps if the off axis response is similar to the on-axis response. I use constant directivity waveguides. I think Dr. Toole, Dr. Olive at Harman have done a wonderful job with the spinorama system and subjective listening tests, with the ability to predict a good sounding loudspeaker based on the anechoic data.

- below Schroeder the room comes into play and could use some assistance. Harman and several other vendors offer hardware and software solutions to assist. Some do it better than others.

- time alignment/time domain correction - I wish there was more research performed in this area. In my limited subjective listening experiments in-room, I can create a filter that has identical frequency response, but one filter is time domain corrected and the other not. In a controlled experiment, I can hear a subtle difference and reference that in my latest article on CA. I am humble enough to admit that I could be "hearing things" and hence would love to see more research. But my results seem to correlate with others that have performed similar experiments. Also note all of my experiments are using linear phase digital XO's. There are no passive XO's in the experiments I performed.

- eq above Shroeder. This is a possible area of divergence. As Dr. Toole points out, a well designed loudspeaker should not require eq above Schroeder. If I had Salon2's or JBL M2's in my living room, it is quite likely that I would not use any eq, based on the spins I have seen. But this is the crux of the issue Cosmik. Really well designed loudspeakers like the Revel's and JBL's and likely the D&D 8C's are far and few between. The vast majority are not well designed. Some have horrible off-axis response (i.e. uneven directivity) and for sure eq can't fix that. That's why I use constant directivity waveguides. Poorly designed passive XO's that produce lobing, again can't be fixed with eq. Hence why I use linear phase digital XO's.

But here is the biggy - the vast majority of speakers to my ears are (way) too bright sounding and directly correlates to the in-room steady state measurements I have taken. One typical example is the Dynaudio 600 XD's If you look you can see the in-room measurement compared to my reference is some 5 dB hotter from about 3 kHz on up. Way too bright for my ears. In fact, I can't even listen to it. So what to do? Unfortunately, shelving tone controls don't work as what one needs is a tilting tone control. And if there were such a thing, I could probably be happy. But since I am already using DSP to assist below Schroeder, I might as well use it for a tilting tone control. Which is what it is doing above Schroeder as it is mostly the direct sound that is being eq'd. In combination with linear phase digital XO's and constant directivity waveguides, the off axis response is almost identical as the on-axis response. I show that in detail my book.

Cosmik, you can pick this apart, but I hope you can see that while the approach is different, the end result is similar. I can't afford Salon 2's or M2's at the moment, but I can afford some sophisticated DSP that assists in smoothing out the bottom end and eq's the tonal response (i.e. direct sound timbre) to something that I prefer without having my ears ripped off from too much high frequency energy. As mentioned, I am hopeful that other speaker manufacturers start leveraging the great research from Dr. Toole and Dr. Olive and start producing better sounding speakers.
Thanks for the lengthy reply.

...a well designed loudspeaker should not require eq above Schroeder
I would like to know your justification for saying this. I agree with it totally, but find it very hard to persuade anybody else that it is true!

...the vast majority of speakers to my ears are (way) too bright sounding
My first attempt at speaker building indeed had this characteristic - I was naively considering only the on-axis anechoic response and the speaker didn't have perfect dispersion characteristics. Extremely detailed, but horrible to listen to after a while.

The total, 100%, first time solution to the problem was 'baffle step compensation' by calculation. The curve was calculated, and all I did was set the depth of it by ear. It doesn't sound like "turning down the treble" or "turning up the bass", but more like "turning up the richness", and I doubt that any other approach (fiddling manually with a graphic equaliser or inverting in some way a measured in-room curve) could have achieved the integrity and 'objectivity' of that smooth curve.

I think there may be more mileage in this than the in-room measurements approach: get the speaker's time domain performance and basic on-axis EQ correct, then predict the dispersion defects of the speaker based on its 'geometry' and apply formulas that logically and rationally partially compensate for the defects - using ears to set the depth, but calculation to create the curve.

And preferably start with a reasonable speaker of course!
 

Fitzcaraldo215

Major Contributor
Joined
Mar 4, 2016
Messages
1,417
Likes
474
#67
So if I decide to set Audyssey editor to not eq above X freq, where would you guys recommend I set it then, around 200 khz?
Dr. Toole, I believe, recommended about 500Hz, and, as we know, he has excellent speakers. But, some experimentation and subjective listening are highly recommended, as with any attempts at EQ.
 

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
1,946
Likes
1,342
Location
Monument, CO
#68
So if I decide to set Audyssey editor to not eq above X freq, where would you guys recommend I set it then, around 200 khz?
Depends on the room and the speakers but somewhere in the 200 to 500 Hz range would be my guess. Above any strong bass modes/reflections but below the main direct sound contribution.

200 kHz would be a little high for me. ;)
 

Sal1950

Major Contributor
The Chicago Crusher
Joined
Mar 1, 2016
Messages
4,125
Likes
1,235
Location
Central Fl
#69
Thanks for the response guys, just pickin your brains for ideas. It's easy enough for me to change, there's just a slider in the new Audyssey Editor app that I can put anywhere I like and reload the file to the Pre/pro. Guess I just need to try a few different points and listen. From reading thru this thread it seems more a subjective preference between ending DSP somewhere in the Shroeder area or having the slowly desending curve vs flat.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
5,362
Likes
1,333
Location
Riverview, Fl
#70
Thanks for the response guys, just pickin your brains for ideas.
Show an uncorrected sweep and maybe that will yield a clue...

Mine, 1/3 smoothing (uncorrected, both L/R speakers active):

Blue - JBL LSR 308
Red - Martin Logan reQuest

upload_2018-5-13_16-13-25.png


If I had to guess, 250Hz would be a reasonable endpoint, here.

They're both a little heavy on the lows to my ear, at higher moderate volumes.

1/48 smoothing:

upload_2018-5-13_16-17-41.png
 
Last edited:

Fitzcaraldo215

Major Contributor
Joined
Mar 4, 2016
Messages
1,417
Likes
474
#71
Thanks for the response guys, just pickin your brains for ideas. It's easy enough for me to change, there's just a slider in the new Audyssey Editor app that I can put anywhere I like and reload the file to the Pre/pro. Guess I just need to try a few different points and listen. From reading thru this thread it seems more a subjective preference between ending DSP somewhere in the Shroeder area or having the slowly desending curve vs flat.
Yup. Good, except it's pretty much a done deal theoretically that if you are going to EQ full range - above Schroeder - the gently downward sloping curve with a bit more slope above 15K is the one to use in far field listening. The flat curve is only for near field listening in theory. But, by all means experiment to find the slopes you subjectively prefer across a wide range of recordings.

And, hopefully, the new Audyssey editor will let you remove that damned midrange dip of theirs. I believe that dip may be there even in their "flat" or sometimes grossly misnamed "music" setting, unless it is explicitly removed. However, if for whatever reason you prefer that, leave it in.

Also, even below Scroeder, the stock Audyssey curve as I recall was flat. You might find it advantageous to introduce some slope there as well, tilting up the deepest bass a bit. Consult Toole, Olive and even the old B&K curve on this. And, Audyssey Dynamic EQ - their Fletcher-Munson style loudness compensation - is useless with music, since there are no agreed upon level standards in music recording, unlike movies. But, I did not like Dynamic EQ with movies either.

In my pre-Dirac Audyssey Pro days and with Martin-Logan 'stat hybrids, I was fairly happy with full range EQ using their default "small room volume" downward sloping curve as long as the infernal midrange dip was eliminated. But, I am even much happier using the PC version of Dirac in 5/7.1, since all my media are now sourced from my PC. That, and PC Dirac does no secret, hidden, unspecified downsampling to 48k, unlike most extant versions of Audyssey on HT processors, even those purporting to be 192k capable. Some HT processor versions of Dirac downsample to 48k as well. Others, like Arcam, use 96k. I listen primarily to DSD converted to 176k PCM using PC Dirac.
 
Last edited:

Sal1950

Major Contributor
The Chicago Crusher
Joined
Mar 1, 2016
Messages
4,125
Likes
1,235
Location
Central Fl
#72
And, hopefully, the new Audyssey editor will let you remove that damned midrange dip of theirs. I believe that dip may be there even in their "flat" or sometimes grossly misnamed "music" setting, unless it is explicitly removed. However, if for whatever reason you prefer that, leave it in.
Yes it is individually switchable on all channels.
Messing with this stuff and then trying to use REW to measure results makes my head spin with the multi channel rig. It's all so easy with plain ole stereo comparatively. :confused:
 

Fitzcaraldo215

Major Contributor
Joined
Mar 4, 2016
Messages
1,417
Likes
474
#73
Yes it is individually switchable on all channels.
Messing with this stuff and then trying to use REW to measure results makes my head spin with the multi channel rig. It's all so easy with plain ole stereo comparatively. :confused:
Good luck. The Audyssey calibration is easy enough. It's the followup measurement that is going to be messy.

I admit, I don't use REW or do followup measurements. I just listen and vary the filter calculations, choosing by ear with music over time. And, as we have discussed before, there is an inherent mismatch between the multipoint Audyssey calibration and measurement vs. any single point measurement post-calibration. Which is right?

Also, as you probably know, use a mic floor stand with boom to swing over chairs, and turn off HVAC or other noise sources during the calibration. I even take myself out of the room behind a closed door during the test tones, going back in just to move the mic.
 

Sal1950

Major Contributor
The Chicago Crusher
Joined
Mar 1, 2016
Messages
4,125
Likes
1,235
Location
Central Fl
#74
Options, are they a blessing or a curse?

Good luck. The Audyssey calibration is easy enough. It's the followup measurement that is going to be messy.
I admit, I don't use REW or do followup measurements. I just listen and vary the filter calculations, choosing by ear with music over time.
Yep, with multch it's some type of crap shoot on how you want the the processor to handle the REW sweep signal? There are literally a mountain of different ways you could approach it.
I believe you may be right to just use the ear and preference. Possible is to maybe just REW measure with processor set to "stereo", the DSP and bass management active. Then tweak Audyssey for the desired curve multch wide?

And, as we have discussed before, there is an inherent mismatch between the multipoint Audyssey calibration and measurement vs. any single point measurement post-calibration. Which is right?
I've got no idea. Not possible calculate for a single point measurement, but can let it do the final calculations at any time after 3. I wonder how different from a single point the final curve would be if you just left the mic in the MLP for the 3 runs and then ran the calculation?
Just tooo much to think about and time to spend.
I'm not the kind that finds this fun.
Maybe Ray would be interested in a nice steak dinner? :D
Thanks Fitz
 

Fitzcaraldo215

Major Contributor
Joined
Mar 4, 2016
Messages
1,417
Likes
474
#76
Options, are they a blessing or a curse?


Yep, with multch it's some type of crap shoot on how you want the the processor to handle the REW sweep signal? There are literally a mountain of different ways you could approach it.
I believe you may be right to just use the ear and preference. Possible is to maybe just REW measure with processor set to "stereo", the DSP and bass management active. Then tweak Audyssey for the desired curve multch wide?


I've got no idea. Not possible calculate for a single point measurement, but can let it do the final calculations at any time after 3. I wonder how different from a single point the final curve would be if you just left the mic in the MLP for the 3 runs and then ran the calculation?
Just tooo much to think about and time to spend.
I'm not the kind that finds this fun.
Maybe Ray would be interested in a nice steak dinner? :D
Thanks Fitz
I think one just goes totally with the calibration tool's philosophy - single point or multipoint and stops right there. Mixing the two, as in multipoint calibration with single point post measurement, may risk spurious or unnecessary questions and complications trying to reconcile the two. Believe me, I have followed many Audyssey forums that got lost in doing exactly that. Doing so takes you down the road of assuming the single point measurements are "right" and the multipoint measurements are "wrong". Depending on circumstances and assumptions, both are compromises based only on a limited sample of room issues and neither is right or wrong. And, room EQ is no panacea. It might not solve all measured problems. It seeks improvement, not perfection.

But, you have bought a multipoint calibration tool. Do you not trust it? If not, why did you buy it and why use it at all? And, if you are going to use it, use it as the manufacturer recommends, at least for starters. Then, verify subjectively by ear. Do that double blind if you wish. Adjust the target curve or other parameters like upper EQ cutoff frequency for maximum subjective satisfaction. Tweak it all you want until you are satisfied. The novelty of tweaking the tool gradually lost interest for me. I became more interested spending my time in just listening to music

Yes, one very nice aspect of single point tools is that you can just leave the mike stationary and measure the calibrated results all you want. For a small listening area, that may probably be good enough. If multipoint is better in other circumstances, then the measurement difficulties multiply. But, either should ultimately be capable of delivering a "better" subjective listening experience, even if one that does not measure perfectly. If not, scrap the whole idea.
 

Fitzcaraldo215

Major Contributor
Joined
Mar 4, 2016
Messages
1,417
Likes
474
#77
Can't seem to shut off Audyssey and still have subs active.
This is the current active curve.
View attachment 12700
There should be a way to just turn Audyssey off leaving Bass Management on. I have not seen it otherwise. You also need that for any comparative listening.

The curves are nice, but perhaps overly smoothed in the bass. EQ should do a nice job, we hope.

I assume these are L-R curves. The nice thing potentially about full range EQ is that it might bring the two curves closer together in the mid- and hi-frequencies.
 

jhaider

Active Member
Joined
Jun 5, 2016
Messages
171
Likes
71
#78
So if I decide to set Audyssey editor to not eq above X freq, where would you guys recommend I set it then, around 200 khz?
Depends on your room size. Based on the graph above, 500Hz sounds like a good starting point to me.

And, as we have discussed before, there is an inherent mismatch between the multipoint Audyssey calibration and measurement vs. any single point measurement post-calibration.
Perhaps true as to a single point, I've measured XT32 on two pieces of hardware using a spatial average around the listening position. I made no attempt to match the Audyssey microphone locations, or to use the same locations both times. Yet in both cases the measured results very closely tracked Audyssey's known target curve. Curves actually (Reference, Flat, Bypass LR).

If that hadn't been the case, I think that would have led to the inference that Audyssey is really bad software.
 

Sal1950

Major Contributor
The Chicago Crusher
Joined
Mar 1, 2016
Messages
4,125
Likes
1,235
Location
Central Fl
#79
I think one just goes totally with the calibration tool's philosophy - single point or multipoint and stops right there.
Yes, that only makes sense.
The novelty of tweaking the tool gradually lost interest for me. I became more interested spending my time in just listening to music
I burned out on it 25+ years ago when I first started adding subwoofing to my rig using crude SPL meter methods only. Never fooled much with the last processor and the older XT software, just basicly ran the wizard and was done with it since there wasn't much you could tinker with anyhow. Now got this new kit with XT32 and the add-on Editor software and got suckered back in. LOL I can only fool around for an afternoon, then I have to walk away for a couple days before I look at it again.
 
Joined
Mar 17, 2016
Messages
46
Likes
19
#80
Just implemented this. Works great for music but may need some tuning for HT (This is the B&K curve in case you are wondering). While it looks steeper than it is, it is only about a 5dB drop from 50Hz to 20K Hz./ Depending on the high end of your speakers, that portion of the target may need diddling/adjustment.
B&K.png
 

Similar threads

Top Bottom