Fluid is right about the room gain, you still get the same amount of room gain with arrays. If that weren't the case, I could not have run my arrays full range. They would not have the required volume displacement without the added room gain to do the bottom end justice.
A single driver would have floor bounce as seen from a listeners perspective. This usually leads to an interference dip as fluid mentioned. Double up on mid-bass drivers and that dip gets less deep and spread out over a wider part of the frequency range.
Get the pair close to the floor and it would look like a larger stack (or array) of drivers from the listeners point of view.
An array of drivers still has floor bounce, however this time it is from all the drivers in that array, all with slightly different timing/delay. Each of the drivers only does a small part to create the total desired SPL level, so the bounces are less strong each, and are way harder to trace back in measurements. In a way it looks/acts like a longer continuous source of drivers. Which is why we build these things from floor to ceiling, or at least covering about ~70-80% of that height. We want to mimic an infinite array.
So they don't really avoid the floor and ceiling, their (multiple) floor and ceiling bounce is just way less detrimental to the resulting frequency response. Making it easier to have them work with the room.
A single driver would have floor bounce as seen from a listeners perspective. This usually leads to an interference dip as fluid mentioned. Double up on mid-bass drivers and that dip gets less deep and spread out over a wider part of the frequency range.
Get the pair close to the floor and it would look like a larger stack (or array) of drivers from the listeners point of view.
An array of drivers still has floor bounce, however this time it is from all the drivers in that array, all with slightly different timing/delay. Each of the drivers only does a small part to create the total desired SPL level, so the bounces are less strong each, and are way harder to trace back in measurements. In a way it looks/acts like a longer continuous source of drivers. Which is why we build these things from floor to ceiling, or at least covering about ~70-80% of that height. We want to mimic an infinite array.
So they don't really avoid the floor and ceiling, their (multiple) floor and ceiling bounce is just way less detrimental to the resulting frequency response. Making it easier to have them work with the room.
About the EQ, aside from the fact that it is mandatory for arrays...
In an ideal world, one would create a speaker with a flat response under anechoic conditions and with constant directivity. Take that speaker into a good enough room and it will exhibit an in room response like Toole showed us.
I basically reversed that scenario, I did not look at the speaker response under anechoic conditions. Only taking measurements right there at the listening spot. Admittedly, it's frequency response isn't entirely constant directivity either. However it does have some interesting features, like the 3 dB fall off for each doubling of distance, no crossovers needed and being based on 3.5" drivers it still has quite reasonable off axis plots out to about plus/minus ~20 degree. So if I absorb the side wall reflections, the part that would otherwise create an uneven power response, it's behaviour is getting pretty good over a reasonably wide area.
To decide upon size and especially the thickness for my damping panels I reviewed the in room measurements to see where the room had uneven response in the reverberant plots, in other words uneven and excessive output from the room, caused by the off axis response of the drivers I used.
Goal of those absorbing panels: absorb first reflections and create an even room response, much like it would have been if the speaker itself had a better power response. It cleaned up the direct sound/reflected sound and made the indoor plots look great
.
Next step was to find an EQ curve that worked in my room. Even though I did try some known room curves, like the B&K curve, another curve from Mitch/Bob Katz, I also tried what Siegfried Linkwitz proposed, I did not find something that simply worked well for all music I liked.
I picked the one that worked best (which was another JBL published curve, actually) for most recordings and started to adjust bits and pieces of it over time. This way, I was gradually moving towards a curve that worked well in my room.
Over this same period I started to work on the mid/side EQ, further obscuring the real or actual curve I was listening to. Making it even harder to see what worked.
It was only when I started to look at the side part by itself (from that mid/side EQ), the curve used for the sounds that are panned hard left or right, without the tweaks I used within my mid/side EQ, that I first noticed the shape of that curve had a large resemblance to the JBL preferred listeners curve. I've been promoting it ever since and now use it as a sort of guide line. I may still deviate somewhat, the biggest deviation from it would be the mid/side EQ tweaks.
In an ideal world, one would create a speaker with a flat response under anechoic conditions and with constant directivity. Take that speaker into a good enough room and it will exhibit an in room response like Toole showed us.
I basically reversed that scenario, I did not look at the speaker response under anechoic conditions. Only taking measurements right there at the listening spot. Admittedly, it's frequency response isn't entirely constant directivity either. However it does have some interesting features, like the 3 dB fall off for each doubling of distance, no crossovers needed and being based on 3.5" drivers it still has quite reasonable off axis plots out to about plus/minus ~20 degree. So if I absorb the side wall reflections, the part that would otherwise create an uneven power response, it's behaviour is getting pretty good over a reasonably wide area.
To decide upon size and especially the thickness for my damping panels I reviewed the in room measurements to see where the room had uneven response in the reverberant plots, in other words uneven and excessive output from the room, caused by the off axis response of the drivers I used.
Goal of those absorbing panels: absorb first reflections and create an even room response, much like it would have been if the speaker itself had a better power response. It cleaned up the direct sound/reflected sound and made the indoor plots look great
.Next step was to find an EQ curve that worked in my room. Even though I did try some known room curves, like the B&K curve, another curve from Mitch/Bob Katz, I also tried what Siegfried Linkwitz proposed, I did not find something that simply worked well for all music I liked.
I picked the one that worked best (which was another JBL published curve, actually) for most recordings and started to adjust bits and pieces of it over time. This way, I was gradually moving towards a curve that worked well in my room.
Over this same period I started to work on the mid/side EQ, further obscuring the real or actual curve I was listening to. Making it even harder to see what worked.
It was only when I started to look at the side part by itself (from that mid/side EQ), the curve used for the sounds that are panned hard left or right, without the tweaks I used within my mid/side EQ, that I first noticed the shape of that curve had a large resemblance to the JBL preferred listeners curve. I've been promoting it ever since and now use it as a sort of guide line. I may still deviate somewhat, the biggest deviation from it would be the mid/side EQ tweaks.
Last edited:
Good to hear your voice again fluid
+1
Thanks Guys
Any free time I get these days goes to house renovations or landscaping. My son is also at the destroy / climb everything stage so the arrays are safely locked away for now the down side is that they are not set up for use either
I see one curve were you have a 5dB shelf below 100 Hz, then response drops off by ~5dB from 100 Hz to 16 kHz. An Olive slope with a bass shelf. Is that where you are now?
This is the curve I have, aside from the mid/side EQ tweaks:
It's closest to the "Trained listeners curve" as presented by JBL.
Not too far off from earlier examples, but minor tweaks do have perceptual differences, if they span a couple of octaves.
The curve you mention has been used in the past, but on top of that I tuned the mid/side EQ. The above curve lies in the middle of all mid/side tweaks.
It's closest to the "Trained listeners curve" as presented by JBL.
Not too far off from earlier examples, but minor tweaks do have perceptual differences, if they span a couple of octaves.
The curve you mention has been used in the past, but on top of that I tuned the mid/side EQ. The above curve lies in the middle of all mid/side tweaks.
This graph shows what happens when mid/side EQ is included:
The dark blue line, following the JBL curve, is what you get for sounds panned hard left or hard right.
The upper green line would only be valid for L-R or R-L signals. The line with the dips at ~3.4 KHz and 600 Hz is the phantom trace.
Small tweaks can still further evolve these results. Overall this changes the perceived balance that makes the side panned sounds and phantom sounds sound more alike. It gives the side panned sound way more body and removes a bit of a boxy sound of the phantom center. It makes the center panned sound more intelligible, easier to follow and makes it stand out more. There's still more experimenting to do, I'm sure. Just reporting what worked for me so far. It works in conjunction with the virtual reflections and tweaking the balance between mid and side has a huge effect too.
Levels of early reflections out in the room play a huge role in perception here. I hardly have any loud early reflections present and introduce late reflections (after ~15 to 17 ms) which further help the perceived balance. Basically it fills in the tonal gap we get due to our two ears at ~1750 kHz in the exact sweet spot. The virtually introduced late reflections are split into mid and side too, with each having a slightly different balance and crossover point. The sides, consisting of L-R or R-L, are pretty flat and have a 12 dB/oct crossover set at 3.5 KHz, the phantom part is running up higher to 7.5 KHz, it also has some minor tweaks that I'm still working on perfecting.
No idea if this info helps anyone. I still mean to measure at ear positions with something resembling a dummy head one day soon to see if I can perfect this part of the setup. This part of my total chain is causing quite the perceptual change. Making things come alive. A good side effect is the reduction/removal of perceived sibilance in phantom sounds. It works wonders to create a very live like 3D type of imaging.
For mono setups I bet the JPL Trained listeners curve would work very well. All tweaks here are trying to get tonal balance of phantom perceived sounds closer to what the sides sound like. If one splits the stereo into mid and side, the mid part still contains side info (L+R). This makes it a bit harder to figure out what really happens with all this processing.
To get the above graphs I made several IR's to feed trough JRiver:
-A pure left sound, blank/silent right channel
-A pure right sound, blank/silent left channel
-Left plus right
-Left minus right
The dark blue line, following the JBL curve, is what you get for sounds panned hard left or hard right.
The upper green line would only be valid for L-R or R-L signals. The line with the dips at ~3.4 KHz and 600 Hz is the phantom trace.
Small tweaks can still further evolve these results. Overall this changes the perceived balance that makes the side panned sounds and phantom sounds sound more alike. It gives the side panned sound way more body and removes a bit of a boxy sound of the phantom center. It makes the center panned sound more intelligible, easier to follow and makes it stand out more. There's still more experimenting to do, I'm sure. Just reporting what worked for me so far. It works in conjunction with the virtual reflections and tweaking the balance between mid and side has a huge effect too.
Levels of early reflections out in the room play a huge role in perception here. I hardly have any loud early reflections present and introduce late reflections (after ~15 to 17 ms) which further help the perceived balance. Basically it fills in the tonal gap we get due to our two ears at ~1750 kHz in the exact sweet spot. The virtually introduced late reflections are split into mid and side too, with each having a slightly different balance and crossover point. The sides, consisting of L-R or R-L, are pretty flat and have a 12 dB/oct crossover set at 3.5 KHz, the phantom part is running up higher to 7.5 KHz, it also has some minor tweaks that I'm still working on perfecting.
No idea if this info helps anyone. I still mean to measure at ear positions with something resembling a dummy head one day soon to see if I can perfect this part of the setup. This part of my total chain is causing quite the perceptual change. Making things come alive. A good side effect is the reduction/removal of perceived sibilance in phantom sounds. It works wonders to create a very live like 3D type of imaging.
For mono setups I bet the JPL Trained listeners curve would work very well. All tweaks here are trying to get tonal balance of phantom perceived sounds closer to what the sides sound like. If one splits the stereo into mid and side, the mid part still contains side info (L+R). This makes it a bit harder to figure out what really happens with all this processing.
To get the above graphs I made several IR's to feed trough JRiver:
-A pure left sound, blank/silent right channel
-A pure right sound, blank/silent left channel
-Left plus right
-Left minus right
Last edited:
Thanks for that. To me it still looks basically lie a shelf below 100, and a gentle slope down above that. The M/S EQ aside. I don't think I've ever used a curve quite like that, so it would be interesting to try. I've tended to do flat bass and put the downward hinge point at 400 Hz or 1 kHz. Will have to give it a try.
In live sound there tends to be a similar bass shelf, but of greater magnitude. Makes the system sound big and full.
In live sound there tends to be a similar bass shelf, but of greater magnitude. Makes the system sound big and full.
Music is indeed in the ears of the beholder.
My house curve is less pronounced. I tried a 10dB tilt, but it was way too dark for me. I have something closer to 5dB tilt, plus, I like a soft extra couple of dBs between 4k and 8k. To my ears, it gives a little extra clarity to vocals and percussive sounds.
Plus, my room being all concrete reflects a lot of energy, so I have to ease off a bit under 60Hz... otherwise it becomes overwhelming and drowns everything else... even i the FR curve looks perfect, I have to lower the energy that accumulates between 35 and 60Hz.
My house curve is less pronounced. I tried a 10dB tilt, but it was way too dark for me. I have something closer to 5dB tilt, plus, I like a soft extra couple of dBs between 4k and 8k. To my ears, it gives a little extra clarity to vocals and percussive sounds.
Plus, my room being all concrete reflects a lot of energy, so I have to ease off a bit under 60Hz... otherwise it becomes overwhelming and drowns everything else... even i the FR curve looks perfect, I have to lower the energy that accumulates between 35 and 60Hz.
Hi perceval,
Say we all dial in exactly same JBL target curve think theoretical directivity/power responses below for tweeter diameters dictate a few adjustments is needed in the ears of the beholder to sound close to the same in a same room. BTW diameters is set to represent your W8 then TC9/10F size and a real tweeter, at 7kHz 170mm diameter is dived down at a 23º coverage 68mm diameter at 61º and tweeter spray 180º around. Suggest if W8 end too dark give it a try use pure linear phase EQ to counter for sensed region differences between a 25mm and 170mm tweeter because it will ensure timing of harmonics stays same only virtual lifting power response a bit hotter, say a high shelve at 1600Hz Q0,5-0,6 +1-6dB. That said directivity can also be a good thing, measurements will probably tell some : )
Say we all dial in exactly same JBL target curve think theoretical directivity/power responses below for tweeter diameters dictate a few adjustments is needed in the ears of the beholder to sound close to the same in a same room. BTW diameters is set to represent your W8 then TC9/10F size and a real tweeter, at 7kHz 170mm diameter is dived down at a 23º coverage 68mm diameter at 61º and tweeter spray 180º around. Suggest if W8 end too dark give it a try use pure linear phase EQ to counter for sensed region differences between a 25mm and 170mm tweeter because it will ensure timing of harmonics stays same only virtual lifting power response a bit hotter, say a high shelve at 1600Hz Q0,5-0,6 +1-6dB. That said directivity can also be a good thing, measurements will probably tell some : )
Attachments
Last edited:
Think get what you try to tell but how to know what wesayso actual have dialed in on his shared visuals is dark and inline with what you think is represented by your blue curve ? That delta is some reason i pointed to tweeter diameters and calling for other variations also because think if you sat listening in wesayso's couch then dark would never cross your mind . Overall because of small or big system variances think we could discuss about preferred house curve subject relative to each other to eternity if we not have locked close to exactly same acoustic environment and system dialed in documented to exactly same curve.
. Overall because of small or big system variances think we could discuss about preferred house curve subject relative to each other to eternity if we not have locked close to exactly same acoustic environment and system dialed in documented to exactly same curve.
Last edited:
And not the same ears!
Got it.
Yes, I'm sure my big 8" full rangers are beaming a lot more than wesayso's little TC9s. Compensation will be different.
I never had much luck with target freq responses (too "dull" to "heavy" etc...) until I changed my EQ curve from the inversion of a 4 cycles prefiltered response, to the inversion of the freq response envelope of a longer (8-12 cycles) prefiltered response. Now I seem to do well with a downward "tilt" on a linear freq scale (attenuation doubles with each octave increase). It's fairly close to the Bob Katz "knee" target. I'm going to be moving to a new house in the near future; it will be interesting to see how this method translates to a different room.
Perceval, just looking at the JBL plot already shows a wide variety of preferences.
The room will play a huge role in that preference. Within my room, there isn't much difference between my balance at a frequency dependent plot and the longer windows that include the room. I'm not saying the room does not factor in though, it sure has a stamp of it's own, but the frequency response or balance of the room is mimicking the balance of the direct sound.
You like open baffle, no doubt the room will respond after the direct sound in such a case. This will have a definite influence on the perceived tonal balance.
@Pano, even though it might look like that Sean Olive declining plot with a bass lift, the relative flat spot from 500 to about 1750 Hz is essential for my preferred balance. So you have that bass shelf, then a gradual decline out to 400-500 Hz and relative flat out to 1750, from where it goes down again.
That gradual part from 60 Hz to about 500 Hz is adding just a touch of warmth, use any more and it will 'cloud' or obscure the midrange. In conjunction with mid/side EQ this sounds very 'open' to my ears. It's all about balance of certain parts of the frequency range compared to other parts.
A straight line decline never sounded completely right to me, at any angle. Trust me, I tried many angles on that plot.
However I do need to note that it seemed to work on some material, but never on everything I tried.
The balance I was looking for, or wanted was the one that worked best along or with many different genre's and types of music.
Generally, we are very used to hear the room we are in which always alters our perceived sound/balance. Without thinking about it our brain filters out most of that nasty imbalance and lets you hear a filtered result. Most people simply accept that (certainly nothing wrong with that) and enjoy the end result of their speakers + the added room response.
I wanted something different from the start. I wanted to maximise, wherever possible, the feeling of "being there". This required cleaning up the reflections to get rid of the large hints of the room you're actually in.
The more you clean up that indirect part, the more aware of it you become. Things like the cross talk between both ears becomes more noticeable, there are no reflections left to fill in the dips that the inter aural combing creates. But I heard big differences between recordings too, making me wonder: is it possible that I do not like the recorded room in this song?
This made it increasingly harder to find a satisfying balance, one that fits all. At least, one that works well over a wide selection of (my preferred) music. Each Era in music even had it's own particular 'stamp' or 'added sound signature' on the final mixes.
As headphones usually give quite good results on various material, a good clean room should be able to do the same, I kept repeating to myself.
Easier said than done though, making it clear to me that we usually accept what the room answers quite easily. But after altering that, we become way more aware of what traces or coloration may be left. Over time and after months of adjusting my balance, I got the curve that resembled that JBL plot. I even had it for quite a while before realising that those two (my plot and the one from JBL) were so close. It was literally buried in different parts of EQ, mid/side EQ included. It all was a big mess of EQ settings everywhere, tweaks that were added over time.
It was when I re-engineered my prior results from scratch (from old measurements and JRiver data) to build it up again (after a slight computer mishap) that I first saw the true curve I was listening to.
For a while I expected to end up with more of a straight line. Even doubting my preference for this funny curve. So in a way I was glad to find this funny looking JBL plot. Even slight differences can have quite huge effects in listening. I'm still surprised that for me, the difference between good and great have been minor tweaks.
The room will play a huge role in that preference. Within my room, there isn't much difference between my balance at a frequency dependent plot and the longer windows that include the room. I'm not saying the room does not factor in though, it sure has a stamp of it's own, but the frequency response or balance of the room is mimicking the balance of the direct sound.
You like open baffle, no doubt the room will respond after the direct sound in such a case. This will have a definite influence on the perceived tonal balance.
@Pano, even though it might look like that Sean Olive declining plot with a bass lift, the relative flat spot from 500 to about 1750 Hz is essential for my preferred balance. So you have that bass shelf, then a gradual decline out to 400-500 Hz and relative flat out to 1750, from where it goes down again.
That gradual part from 60 Hz to about 500 Hz is adding just a touch of warmth, use any more and it will 'cloud' or obscure the midrange. In conjunction with mid/side EQ this sounds very 'open' to my ears. It's all about balance of certain parts of the frequency range compared to other parts.
A straight line decline never sounded completely right to me, at any angle. Trust me, I tried many angles on that plot
. However I do need to note that it seemed to work on some material, but never on everything I tried.
The balance I was looking for, or wanted was the one that worked best along or with many different genre's and types of music.
Generally, we are very used to hear the room we are in which always alters our perceived sound/balance. Without thinking about it our brain filters out most of that nasty imbalance and lets you hear a filtered result. Most people simply accept that (certainly nothing wrong with that) and enjoy the end result of their speakers + the added room response.
I wanted something different from the start. I wanted to maximise, wherever possible, the feeling of "being there". This required cleaning up the reflections to get rid of the large hints of the room you're actually in.
The more you clean up that indirect part, the more aware of it you become. Things like the cross talk between both ears becomes more noticeable, there are no reflections left to fill in the dips that the inter aural combing creates. But I heard big differences between recordings too, making me wonder: is it possible that I do not like the recorded room in this song?
This made it increasingly harder to find a satisfying balance, one that fits all. At least, one that works well over a wide selection of (my preferred) music. Each Era in music even had it's own particular 'stamp' or 'added sound signature' on the final mixes.
As headphones usually give quite good results on various material, a good clean room should be able to do the same, I kept repeating to myself.
Easier said than done though, making it clear to me that we usually accept what the room answers quite easily. But after altering that, we become way more aware of what traces or coloration may be left. Over time and after months of adjusting my balance, I got the curve that resembled that JBL plot. I even had it for quite a while before realising that those two (my plot and the one from JBL) were so close. It was literally buried in different parts of EQ, mid/side EQ included. It all was a big mess of EQ settings everywhere, tweaks that were added over time.
It was when I re-engineered my prior results from scratch (from old measurements and JRiver data) to build it up again (after a slight computer mishap) that I first saw the true curve I was listening to.
For a while I expected to end up with more of a straight line. Even doubting my preference for this funny curve. So in a way I was glad to find this funny looking JBL plot. Even slight differences can have quite huge effects in listening. I'm still surprised that for me, the difference between good and great have been minor tweaks.
Nice post. I especially like the following quotes:
"It's all about balance of certain parts of the frequency range compared to other parts."
"Generally, we are very used to hear the room we are in which always alters our perceived sound/balance."
"It's all about balance of certain parts of the frequency range compared to other parts. A straight line decline never sounded completely right to me, at any angle. Trust me, I tried many angles on that plot . However I do need to note that it seemed to work on some material, but never on everything I tried. The balance I was looking for, or wanted was the one that worked best along or with many different genre's and types of music."
"As headphones usually give quite good results on various material, a good clean room should be able to do the same, I kept repeating to myself."
I can relate here especially since I've been trying to develop an automated EQ scheme that works well with all types of recordings and in different rooms. I think I've achieved some success with the first part, but no way for me to test the second part on my own at this point...
Anyway, always fun to read along here.
"It's all about balance of certain parts of the frequency range compared to other parts."
"Generally, we are very used to hear the room we are in which always alters our perceived sound/balance."
"It's all about balance of certain parts of the frequency range compared to other parts. A straight line decline never sounded completely right to me, at any angle. Trust me, I tried many angles on that plot . However I do need to note that it seemed to work on some material, but never on everything I tried. The balance I was looking for, or wanted was the one that worked best along or with many different genre's and types of music."
"As headphones usually give quite good results on various material, a good clean room should be able to do the same, I kept repeating to myself."
I can relate here especially since I've been trying to develop an automated EQ scheme that works well with all types of recordings and in different rooms. I think I've achieved some success with the first part, but no way for me to test the second part on my own at this point...
Anyway, always fun to read along here.