Jump to content
Sign in to follow this  
SME

My living room "make over" (aka the "surrounded by bass" project)

Recommended Posts

So lately, I've probably been enjoying my existing system too much and not spending enough time working on my subs.  I've continued to make very minor tweaks to the system.  The biggest change was to back off on the boost to 190 Hz on the center channel and instead make it ~flat "on-average" across the seats.  This is mostly because I often sit off-axis, and it didn't sound so good in those other spots.  The excess 190 Hz tended to overwhelm and mask out a lot of detail in voices whose fundamentals hit that spot.  It is an acceptable compromise until I extend the baffle.  The center still sounds very good, and it's nice to not have the weird feeling (literally) that arose from the build-up of later arriving energy at that frequency.

 

Otherwise, I boosted my sub bass response a bit, giving it a bit of an upward tilt.  I think the "first arrival determines tonal balance" concept doesn't apply in the same way to bass because room gain tends to naturally boost even first arrival sub bass in most situations.  An anechoically flat speaker will typically run hot in the bass without EQ adjustment, and that's often just the way things are monitored.  Instead, I've opted to adjust things by ear, and a few extra dB with some slope up toward the bottom seems to sound better.  There is a very clear point where too much low end washes out much of the transient detail.  Kick drum might go boom but not really thump, and bass instrument can sometimes go from being punchy and well-defined to indiscernible sludge.  If I raise the response to that threshold and then back-off about 0.5 dB, repeating this process for different regions of the response, I get a very solid, satisfying thump with no overhang from most kick drum.

 

Where I've struggled more is with taming the high end.  Running the top end flat just doesn't work for my ears and for most others.  In fact, it can be very irritating with a lot of content that I think most would regard as being well recorded.  What I'm finding is that the shape of the roll-off is to an extent more important than the amount of roll-off involved.  Very minute changes to the shape can have a big impact on the sound.  At the same time, content is quite variable in overall top octave strength, so I feel that having an adjustable UHF control will be advantages. 

 

I recently decided to experiment with a hypothesis that the best roll-off shape will almost approximate the roll-off caused by dissipation of sound as it moves through the air.  I was surprised to see that dissipation is actually pretty significant even over short distances for the highest frequencies.  After reviewing data of UHF roll-off vs. distance at various humidity levels, I settled on simulating distance roll-off via a single biquad centered at 12750-13000 Hz (higher for greater roll-off amounts) with a Q of 0.5 and gain ranging from 0 to -10 or so, approximately equivalent to roll-offs due to distances ranging from 0 to 20 meters or so.  With most content, I seem to do well by about -3 dB or ~6 meters of distance and occasionally opt to push it out to -6 or -10 dB for ~12 meter or ~20 meters of distance.  The consequence of adjusting the roll-off while maintaining the constraint of physical realism is is remarkably subtle.  It kind of acts like the audible equivalent of a sharpness control.  Too much, and the transients have a bit too much bite.  Too little, and there is a loss of fine detail as well as crispness.  These changes are very subtle, even compared to say boosting 15-20 kHz by 1 dB *without* altering the frequencies below and thus ending up with a physically unrealistic curve.

 

Along these lines, it's very fascinating to me that UHF and very low frequencies can mask one another substantially.  For a few hours, I accidentally had 15-20 kHz and up running about 1 dB hotter than I meant to run it, and it caused much of the sub bass to sound very distant and lose almost all of its weight and feeling.  But it only happened on certain soundtracks, those with full extension.  I was really weirded out until I recognized that it was the UHF that was masking the bass and causing it to diminish.  I tried pushing the subs up 3 to 6 dB, and it did nothing for its subjective loudness or power and merely muddied the sound.  Without fixing the UHF, there was no helping the bass.  Bringing the UHF to proper balance totally fixed the problem completely to my great relief.  I have also observed masking in the other direction.  When I had some harshness in the UHF, boosting around 45 Hz or so seemed to quell it considerably, even when I couldn't conciously hear any content down there.  Psychoacoustics are very weird.

 

Another thing I encountered when playing with UHF response and ending up too hot in parts was a sort of unrealistic hyper-detailed sound.  I haven't seen a Darbee before, but I imagine it to be the audio equivalent of turning the Darbee effect up way too high.  Upon first impression, the sound is incredibly enhanced, with far more richness and detail than is heard normally, but only upon more careful disciplined listening does it become clear that the sound is completely inaccurate and unrealistic.

 

At this point, I am declaring this experiment a success.  The sound seems to have improved by another notch.  Cymbals and hats sound absolutely great.  Indeed, I'm hearing cymbals emerge in background content of soundtracks where I never heard them before.  And there's even less harshness than there was before.  For the occasional soundtrack that still "bites" a little, I can easily dial-down the UHF without losing detail or weirding out the tonal balance.

Share this post


Link to post
Share on other sites

I tend to agree with much of the above. I tried a ruler flat on axis response outdoors and at the LP in smaller rooms a few times and neither sounds good to me. The top end is subjectively too much for my tastes. Sounds unnatural and the low end sounds anemic. This was especially true for the LF outdoors. I assume it is from the lack of direct vibrational queues, or other types of coupling with the body directly, that are much more prevalent in a vehicle or small room, that might help fill in the LF experience. Overall I prefer a bit of a downward tilt from the bass to the treble range, with the bass becoming a bit more aggressively boosted below 100hz. Most of my friends who are not very technically savvy when it comes to acoustics or speakers seem to agree with that type of general shape as well. Maybe it's technically a coloration preference but whatever. At the end of the day it is all about enjoying your music or movies isn't it?

Share this post


Link to post
Share on other sites

Good posts :)

 

I smiled at the first part, though - it does seem as though sometimes the tweaking time overtakes the actual time spent just playing content for fun!

 

I tend to like the presentation of mine, but I'm just running Audyssey for processing.  Perhaps one gets to like what one is used to until one is presented with something that is better?

 

I do take my hat off to everyone who spends so much time improving their system - I've still not got round to putting the MiniDSP (which I bought about 18 months ago) into the system, and I really must do it soon before I have to move out and back into the missus' dad's house so we can save for a bigger place...

Share this post


Link to post
Share on other sites

I added a discussion along with some in-room response measurements of my system, as it was last time I measured, to the first post of this thread.  (Scroll down to the second section.)

 

I tend to agree with much of the above. I tried a ruler flat on axis response outdoors and at the LP in smaller rooms a few times and neither sounds good to me. The top end is subjectively too much for my tastes. Sounds unnatural and the low end sounds anemic. This was especially true for the LF outdoors. I assume it is from the lack of direct vibrational queues, or other types of coupling with the body directly, that are much more prevalent in a vehicle or small room, that might help fill in the LF experience. Overall I prefer a bit of a downward tilt from the bass to the treble range, with the bass becoming a bit more aggressively boosted below 100hz. Most of my friends who are not very technically savvy when it comes to acoustics or speakers seem to agree with that type of general shape as well. Maybe it's technically a coloration preference but whatever. At the end of the day it is all about enjoying your music or movies isn't it?

 

As per the discussion in my first post, there's a world of difference between flat *first arrival* frequency response and flat frequency response in a typical listening room.  I'm fairly certain that this is why a frequency response with significant slope sounds better indoors.  As for outdoors, I would expect a flatter looking response to sound better, albeit with some high frequency adjustments and some bass boost.  I already discussed the reasoning for high frequency adjustments in my first post.  I did not discuss the bass boost however.

 

I believe I've stated this elsewhere before, but I believe bass boost may actually be more correct.  The reason is that a typical monitor with flat anechoic response will exhibit room gain in a typical listening space, even for the so-called first arrival.  In the early days of audio, this room gain was probably never compensated for in the monitoring systems.  Instead, the room size and speaker placements in the mastering studio were chosen to be approximately representative of listener's homes.  Even today, room gain is likely not compensated for directly.  Instead, a lot of mixers will adjust the bass response of the monitoring system by ear until it sounds good, using trusted recordings for guidance.  In other words, bass boost is established by precedence, just like high frequency adjustments are.

 

When you say that you and most of your friends prefer the response a certain way, it's probably not because you prefer a colored sound.  Sure, there are people that prefer to play their subwoofers as loud as possible so they can experience their eyes wobbling and struggle to breathe.  That's a special case, where extreme bass at the expense of the rest of the sound is enjoyed as sort of a sport.  But when it comes to fully appreciating real world music and movie content, outside of a few dedicated "bass" genres, I believe preference is a lot less subjective than we are commonly led to believe.  The sound that you and your friends prefer to listen to may be a lot closer to what the mix and mastering engineers heard than you would think.

 

 

Good posts :)

 

I smiled at the first part, though - it does seem as though sometimes the tweaking time overtakes the actual time spent just playing content for fun!

 

I tend to like the presentation of mine, but I'm just running Audyssey for processing.  Perhaps one gets to like what one is used to until one is presented with something that is better?

 

I do take my hat off to everyone who spends so much time improving their system - I've still not got round to putting the MiniDSP (which I bought about 18 months ago) into the system, and I really must do it soon before I have to move out and back into the missus' dad's house so we can save for a bigger place...

 

I've put an enormous amount of time into tweaking, and I believe it's really paying off for me now.

 

I liked Audyssey MultEQ XT while I used it, but I now consider it to be seriously flawed.  Unless one's room is acoustically dead, it yields a very top heavy tonal balance.  Admittedly, I didn't realize this is for at least a couple years.  It certainly improves the sound in some aspects, but these improvements come at great cost with respect to other aspects.  Once I got the Pro kit with the adjustable target curve, I figured out how wrong flat was, but unfortunately, the Pro kit was too limited (+/-3 dB max difference in the target curve) and too tedious to use for me to find an optimum target curve.

 

I've heard that XT32 uses a completely different algorithm, but I'd hazard a guess that the result is still very top heavy.  Audyssey makes many claims that their system leads to a "reference" response, implying that a flat frequency response (with a tiny bit of top end roll-off) allows one to hear what the mix/mastering engineers heard in their studios.  Audyssey also claims that if one does not like the flat sound, then he/she has a preference that is inconsistent with the reference response.  These claims are totally wrong, and I'm not sure if anyone at Audyssey actually knows any better.  It's really quite absurd, because despite all of the psycho-acoustic research that supposedly goes into their technology, their approach appears to be based on a flawed understanding of how hearing works and a flawed understanding of how mixes are produced.

 

Definitely get that MiniDSP unit up and running.  Are you going to use it for your mains or just your subs?  Either way, I suggest using the MiniDSP exclusively for EQ and turning off Audyssey.  If your speakers are decent, they should have a much nicer tonal balance than what Audyssey gives you, and if you have EQ capability for them with the MiniDSP, then you can fine tune them as needed.

Share this post


Link to post
Share on other sites

I recently migrated bass management from my AVR to my custom DSP.  This will allow me to do BEQ correctly and to do more powerful room EQ when I get there.

 

I've mentioned in the past that I found calibrating "first arrival" flat sounds best, with some exceptions.  One of those exceptions is the apparent need for a shelf of -1 dB centered in the upper mids for some music.  However, I'm trying a different approach now using a bass boost instead.  The idea is that an anechoic flat speaker has a bass boost from the floor reflection that shows up somewhere in the low mids or upper bass, depending on the design of the speaker and listener distance.  The boost is roughly in the range of 3.5-4.5 dB or so, and I make the center frequency of the boost adjustable to account for different mastering environments.  I'm still doing a lot of experimentation and listening, but moving the center frequency of the boost a bit higher seems to work well (as in sounds good) in lieu of the upper mid shelf.  It's not perfect, but I don't know if it ever will be.  A real floor bounce also has a dip in first arrival sound that appears above where the boost sets in.  Should I simulate the response dip too?  I'd rather not, especially being that its location, shape and depth will vary a lot more in different environments than the boost will.

 

So I'm pretty happy with how my music sounds.  However, I've been recently reminded that cinemas and dub stages are usually calibrated so that in-room power response follows the X curve.  Because of this, some movie content needs the X curve applied to the highs to sound right.  Some also seems to need an extra bit of upper mid cut, perhaps because some room reverb may be present there.  It depends on the room.  Making matters worse is the low frequencies.  In a dub stage with a lot of bass build-up the mains may be calibrated with a lot of attenuation down there in order to keep the power response flat.  As such, the bass boost I use is likely inappropriate for a lot of films.  At least its presence is less likely to be offensive than the lack of X curve where it's needed, but the bass boost may cause some boominess in some of the voices.   To make matters weirder, I actually need that bass boost in my calibration in order for my near-field subs to give me the same SPL as my mains via pink noise.  So is my bass boosted or not?

 

Needless to say, the cinema standard is a disaster for good sound.  The existence of home mixes just throws another wrench into the works.  Do they calibrate to a target curve or do they let the monitors run as they are designed (i.e. flat anechoic)?  Unfortunately, the one example of a home mix I'm aware of is very irritating without a -3 dB high shelf in the upper mids, and the voices are boomy unless I disable my bass boost.  I have no idea how they calibrated their system, but clearly it produced a mix with a very different tonal balance than most music.

 

It's now abundantly clear to me that the only way to get high end sound with a wide variety of movies is to re-EQ them as needed, and not just to restore filtered bass.  Right now, I must make the EQ changes manually, which is disruptive when trying to watch movies, but in the long run, I plan to implement a handful of enlightened tone controls that I can adjust remotely while a movie.  I could use this for music too, where I'd like to be able to adjust bass boost center frequency and UHF/distance roll-off amount more easily as well.

Share this post


Link to post
Share on other sites

I'm long over due for an update here.

Earlier in the summer, I worked on getting my new subs integrated.  That was a big upgrade for sure, but I have also continued to work on the DSP configuration for my speakers to get the best sound out of them.  Over time, I have continued to evolve my calibration/voicing technique.  I am purposely, and rather stubbornly, doing this work entirely in-room, despite the scientific evidence that flat anechoic response, on and off axis, is best.

As detailed in previous posts, I experienced unexpectedly good results by using a 1/3rd octave frequency dependent window (FDW) in order to calibrate for a flat tonal balance.  It's been almost a year since I first stumbled across this technique.  Since then, I have continued to apply this method and have elaborated on it.

The first crucial elaboration was to target a flat 1/3rd octave FDW over a spatial average of measurements instead of a single measurement.  I planned on doing this originally in order to optimize the system for multiple listeners.  However, I have found that the use of the spatial average actually leads to better sound at the MLP than the single measurement did.

The above result is surprising if we assume that a listener hears the direct sound, to the exclusion of all else.  However, it makes a lot of sense if instead a listener hears an average of direct sound and early reflected sound.  My single measurement calibration method brought the tonal balance within +/- 0.5 dB or so of ideal, at least above 1 kHz or so.  That's because the SEOS-15 horn itself exhibits about +/- 0.5 dB variation in direct sound response shape (i.e, between the upper mids and treble) across space.  But while +/- 0.5 dB is very good, relative to most speakers, I discovered I could do much better.

When doing spatial averaged calibration, there is some ambiguity as to the size of the spatial window within which the measurements should fall.  Most speakers exhibit increasing directivity with high frequencies so that the response shape will tend to slant down more to the treble as one moves off-axis.  For this purpose, I settled on a window of approximately +/- 30 degrees, taking some inspiration from the "listening window" used in JBL/Harman's anechoic measurements.  For example, the JBL M2 is EQed to be flat, not on-axis, but within the +/- 30 degree horizontal and +/- 10 degree vertical window.

The second crucial elaboration was to address finer-grained phenomena.  Typically, it is not a good idea to try to EQ out finer-grained peaks and dips in measurements taken in-room, but by EQing a spatial average instead of a single measurement to a target, I can reduce the tendency to EQ room effects that vary by seat and focus more on characteristics of the speaker itself, including resonances in its drivers and power response aberrations caused by cabinet diffraction and close by boundaries that interact relatively uniformly with the speaker.  For this purpose, I continue to rely on the FDW but extend the resolution out to 1/12 octave or so.

For this purpose, it is beneficial to include as many measurements as possible and not just those that fall within the +/- 30 degree listening window.  The idea here is to EQ only the finer-grained peaks and dips that survive spatial averaging and to leave the broad signal shape alone.  There's definitely a bit of push-and-pull here in that higher Q filters still alter the 1/3rd octave FDW response, even though their affect is less than with the lower Q filters.  Calibration of the 1/3rd octave FDW to the target remains the primary objective, not matter what things look like with 1/12 octave FDW, but a smoother 1/12 octave FDW definitely yields substantial benefits.

Only very recently did I tighten down the responses of the left and right speakers enough to achieve a sort of milestone.  I've noticed that as the finer-grained details become cleaner in the upper mids and treble, the sound smooths out immensely.  This time I took a stab at smoothing things more between 200-2000 Hz.  I was reluctant to touch this area because of concerns about EQing the wrong thing, but the spatial averaging seems to help a lot.  Anyway, the result of doing so is a much cleaner and less confused mid-range and even less high frequency harshness.  In fact, one thing I immediately noticed is that I wanted more ultra high frequency (i.e., 8 kHz and up) output.

So the real milestone for me is that I am now running my high frequency response essentially flat all the way to the top, except for a simulated roll-off (about -1.5 dB @ 20 kHz) that accounts for the 3 meters the sound travels through the room to my seats.  I also cleaned up the response a bit more in the ultrasound, and I now extend, flat and smooth in a spatially averaged sense, all the way to 22 kHz.  I expect I will still want the UHF roll-off with some content, particularly where such a roll-off was used on the monitoring system, but thus far, I'm enjoying most music with the flat curve.

I can't emphasize enough how pleased I am with this sound.  It is so clean and smooth.  There is not a hint of harshness with a majority of content.  Percussion sounds very natural.  Shakers, jingle bells, rain sticks, and things like that tended to sound very unnatural without roll-off before, but they sound completely fine now.  I believe I even notice the extra extension to 22 kHz on some recordings, but now it's properly subtle and balanced with the lower frequency content.  Before, it contributed a kind of hyper-real Darbee effect to the sound, and masked all kinds of other content including bass.  Another thing is that the image, depth, and sound-stage are magnificent  and are not confined to the middle seat with stereo listening.

The The last elaboration to comment on concerns low frequencies.  My approach of 1/3rd octave FDW flat works great from 500 Hz on up, but some adjustments are still needed for low frequencies.  I'm not sure why that is and how much the listening window dimensions come into play here, but I suspect that music may simply expect a simple that's not exactly flat below there.  While I've been using a subwoofer bass boost for a long time, the other thing that "just sounds better" is about 1-2 dB dip between roughly 150 and 400 Hz (with more dip closer to 200-250 Hz).  Without this dip, the sound is too muddy with most music.  This may be something that arises naturally (in an approximate sense) when an anechoic flat speaker is placed near a floor in a listening room, due to interference involving the floor.  It's also possible that I would simply benefit from widening the listening window I use for spatial averaging.  I can think of other possible hypotheses, and it's hard to know for certain "what's right" until I can test this approach in a variety of other rooms.  On the plus side, once response at 500 Hz and up is perfected, it's a lot easier (albeit with the practice I have) to EQ the low frequencies by ear.  Of course, there's still benefit to reducing resonances in the low frequencies, which is part of what is being accomplished by EQing modes in the subwoofer range.

Anyway, that pretty much sums up what I've been doing.  On another note, I got a chance to listen to some M2s, and I was quite surprised by how similar I thought the M2s sounded to my speakers.  The biggest differences were in the low frequencies, which are very room-dependent anyway.  My system definitely won on that account.  Curiously, on one of my recent EQ trials, I noticed my speakers sounding even more like I remembered the M2s sounding.  That was before I backed things off just a bit in the UHF.  The M2s do have a rising on-axis response there, and while it sounds kind of cool, it's not really accurate.  Like the M2s, I'm EQing my listening window to be flat, but because the SEOS + DNA-360 has much wider UHF dispersion, the on-axis response doesn't ramp up much at all.

OK.  That's enough for now.  I need to work on tightening the response of my center channel more, and then I can test some movies.  Too bad most cinema mixes are obviously wrong now.  I have a tweak-able X curve pre-set that I use quite a lot now, but it's nowhere near as good as a high quality home mix such as in the recent "Star Wars" movies.  Once I'm satisfied, I'll try to do another round of measurements and post the updated data here in my thread.

So now I can brag about being flat from 5 to 22000 kHz.  Nice.  :)

Share this post


Link to post
Share on other sites
2 hours ago, Infrasonic said:

Yeah but does it sound good?

:P

Haha!  Of course it does, but everyone says that, including me as of a year ago.  ;)  How do I express in words how much *better* it sounds now than it did now?  I can't really.  If I told you that most good stereo recordings project a convincing 3D scene (or rather 2D with depth as height is mostly a mind trick, IMO), what would you think?  I've actually been flirting with the 3D thing for months now, but whereas before it was fleeting and usually required tweaking of the UHF for optimality, it's not now a fairly consistent thing, without tweaks, on soundtracks that aren't too screwed up.

The real question in my mind is: can it get better still?  And by how much?  Every time I think I've nearly perfected things, I discover a lot more performance left on the table.  And   in fact, there is the somewhat paradoxical fact that the cleaner it gets, the more obvious the remaining flaws are, even when they are small.  Fixing the "worst" problem just reveals a new "worst" problem.  I used to think that, say, +/-1 dB at 1/3rd octave resolution sounded "amazing", but now an error of 0.25 dB is like a cardinal sin.  It makes a lot of sense why mix and mastering engineers strive to use the most accurate monitors possible because the minute differences are far easier to perceive.

Edited by SME
fix typo

Share this post


Link to post
Share on other sites

Heheh. Yeah, I know what you mean. I believe you when you say it is better. Just poking fun a lil at your post. You're a very smart guy and knowledgeable but it almost seemed like nothing short of perfection wasn't good enough. Do don't that to yourself. You'll never be happy.... but, it sounds like you are so ignore me. :P 

Sometimes the constant "improving" can have a detrimental effect on the enjoyment of the HT room. I had to learn to love it even with the warts and all!

  • Like 2

Share this post


Link to post
Share on other sites

For me, work and pleasure are very much mixed together.  It's a real joy to hear the improvements at each step, and yes I do enjoy the fruits of my work from time to time.  My wife certainly won't let me get away without doing so.

Want to know something awesome?  I just discovered a glaring flaw in my target curve methodology.  It's not really a new discovery but something I've had at the back of my mind for a while.  The problem is that the short FDW is not flexible with regard to phase shifts, as occurs in the area around crossovers.  I've known about the issue for a while, and I have blamed it for a few issues from time to time.  For example, there's the problem of "disappearing bass", which I can usually link to problems in other areas of the spectrum, particularly around my 950 Hz XO.

So after writing all those positive words over the last few days, I was able to listen to enough stuff to notice that, while I could hear the bass OK, it just didn't seem to have the impact.  And low-and-behold, boosting the sub seemed to have no effect.  Yep.  Houston, we have a problem!

So recognizing that I probably had too much at 1 kHz again, I decided to actually properly quantify the effect of the crossover phase shifts on the 1/3rd octave FDW when applied to a perfectly flat response.  I had to use some external tools to get it right, and I quickly found I had to also had to account for the all-pass filter I use at ~4 kHz to improve integration with the surround speakers (which in turn have an all-pass around 950 Hz).  I also looked at the sub + main crossover, which is a much uglier beast to analyze because of the fact that time-alignment changes a lot from seat-to-seat, but it also essential.

What I found is that not only are the effects of these shifts on the 1/3rd FDW significant, being up to a couple dB in places, but they are they very broad as well.  The effect of the adjacent phase shifts overlap one another.  So basically my whole "1/3rd FDW sets the target curve" theory has been completely misapplied, and I've been a ways away from the "anechoic flat" target.

Anyway, I probably need to eat at least a few crows now, but I'm quite excited at the prospect of getting things dialed in even better.  It does raise some interesting questions though.  My curve up to this point was starting to look a lot like the Harman curve, when viewed in terms of in-room response.  That's not as much the case now, but more on all this later.

I have a busy weekend, and then I have a lot of work to do to set things right.  With the tools I have, I can continue to use REW, but it will be very tedious, especially if I want to take into account the sub XO correctly.  So I am definitely going to have to make my own analytical and visualization tool for this.

Share this post


Link to post
Share on other sites

SME,

How do you spatially average?  I know REW has an average all traces function, but do you utilize that function and take more measurements near MLP to prioritize, or another way?

JSS

Share this post


Link to post
Share on other sites
18 hours ago, maxmercy said:

How do you spatially average?  I know REW has an average all traces function, but do you utilize that function and take more measurements near MLP to prioritize, or another way?

Yeah, I use the REW average all traces function.  Unfortunately though, the average trace is only frequency data.  Ideally, I could create a minimum phase impulse response having the same response.  So when I use the EQ simulator in REW, I have to apply it to a real IR trace and then carefully compare the changes with the average curve.  Mind you, the actual target slopes at the ends because of how the filtering works, so I have to compare that against a known reference (a dirac delta filtered at 1/3rd octave FDW, actually) It's all very tedious.  And now that I realize I need to account for the phase shifts, it's an even uglier process.

Hence, I'll be developing my own software for this in proper time.  I believe my method works, and I believe it's potentially better than any other room/speaker EQ out there.  Even when I was "doing it all wrong" it sounded amazing.  I probably exaggerated a bit when I said I that, anyway.  The overall differences above 200 Hz or so are only +/-1 dB.  After implementing the crude correction (and then "uncorrecting" the bass, which I'd done "right" by ear in the first place), the change in sound is pretty minor.  The difference is more stark in my mind, comparing it to the system I heard last night consisting of M2s calibrated with JBL's new Synthesis processor.  I had previously heard the M2s in that room without room EQ and liked them a lot, even though the low frequencies sounded a bit thin, which I blamed on boundary interference.  This time, I think they were a bit too full, and the subs seemed way too hot.  I couldn't ascertain whether that was done on purpose or not, for the sales benefit, you know.

When I got home, I realized how spoiled I am.

Share this post


Link to post
Share on other sites

I wish REW had a "show all traces" for the Excess Group Delay window.  That way, you know what freq ranges IIR filtering can be used for correction.  Another would be an standard deviation or variance trace under the average trace whenever an average is calculated.

I may need to post this info on the REW forum.

JSS

Share this post


Link to post
Share on other sites

Yeah, I once wished for "excess group delay" in the overlay windows, but I hardly pay attention to that anymore.  I'm almost always looking at things windowed now.  Except for low frequencies, I believe most reflections should not be EQed whether they induce excess delay or not.  That's part of what the spatial averaging helps with.  The reflections will induce peaks and dips that shift with measurement location, so averaging things help remove them so I avoid correcting them.

Share this post


Link to post
Share on other sites

I notice (in that speaker shootout thread on avs) you commented about certain aspects of your room response being SBIR rather than modal in nature. I was curious to know what you meant by this. I suppose I am thinking of 2 specific question. What makes you say it is SBIR rather than modal? what practical difference has this made to your treatment strategy?

Share this post


Link to post
Share on other sites

How do I know certain aspects of my room response are SBIR vs. modal?  I take measurements at multiple spatial locations.  Modes tend to manifest as a peak or dip at approximately the same frequency at all locations.  SBIR peaks and dips tend to shift in frequency as one moves through the space.

As for how that has affected my treatment strategy?  Well, in the subwoofer region it doesn't make a difference at all.  Most practical room treatments don't work well down there anyway.  On the other hand, sub EQ, especially applied to multiple subs independently does work down there.  It works marvelously in fact.  My response at 70 Hz and below is practically anechoic across the listening position by virtue of sub EQ optimizations.

Above the subwoofer region, it doesn't change much for me practically, but it might change things in another room.  Generally, I see it recommended that room modes be treated using substantial absorption in room corners.  That's because many treatments work fairly well in areas of high pressure, and much bass energy tends to congregate there.  I only have one accessible corner in my listening room, and it's already reserved for the Christmas tree.  :)

While treatments in corners should help a lot with high Q / long duration resonances, they don't do a whole lot for SBIR where it's important to absorb as much of the incident sound as possible at the surface that's interfering.  Unfortunately, that often makes it difficult to deal with in a lot of spaces.  Ceiling "clouds" are popular to deal with ceiling problems.  Multiple panels are often used on sidewalls.

The "problems" are by far the worst between 80-200 Hz or so, which is above where the subs can control the sound and below the speakers' baffle steps.  I have a single ceiling panel for my center channel in my room, and plan to install addition panels for my left and right channels once they are installed in-line with the center in the same inverted position. My measurements of the left and right channels suggest that the worst interference occurs due to the the floor-ceiling and ceiling-floor secondary reflections, which have similar path lengths and therefore combine into a single double-strength reflection.  This will change once they are installed in-line with the center.  Reflections from the side walls interfere less than I expected, which is excellent being that one wall has a huge window on it that I don't want to permanently block.

At this point, I'm actually thinking of using 4 of my TD10s in the boxes I plan to build as platforms for my left and right, thus turning those speakers into a kind of pseudo-line array.  Essentially, the horn would cross to the TD12M on top, and then the vertically oriented TD10s below the horn would be brought in somewhere around maybe 500 Hz or so (depending on what I find works best), and then the subs, which are on the floor, coming in somewhere south of 200 Hz.  At that point, the "speaker" would be acting roughly like a floor-to-ceiling line.  The potential benefits of this configuration would be possible reduction of output in the low mid-range and upper bass for seats nearer to the speakers and reduced floor and ceiling interactions overall.  The speakers as they are hold the horizontal pattern quite well down to almost 500 Hz, so ideally I could continue to shadow the near seats in the lower frequencies too.  I really like this concept "on paper" but I haven't done a detailed analysis yet.

Share this post


Link to post
Share on other sites
7 hours ago, SME said:

The "problems" are by far the worst between 80-200 Hz or so, which is above where the subs can control the sound and below the speakers' baffle steps.  I have a single ceiling panel for my center channel in my room, and plan to install addition panels for my left and right channels once they are installed in-line with the center in the same inverted position. My measurements of the left and right channels suggest that the worst interference occurs due to the the floor-ceiling and ceiling-floor secondary reflections, which have similar path lengths and therefore combine into a single double-strength reflection.  This will change once they are installed in-line with the center.  Reflections from the side walls interfere less than I expected, which is excellent being that one wall has a huge window on it that I don't want to permanently block.

At this point, I'm actually thinking of using 4 of my TD10s in the boxes I plan to build as platforms for my left and right, thus turning those speakers into a kind of pseudo-line array.  Essentially, the horn would cross to the TD12M on top, and then the vertically oriented TD10s below the horn would be brought in somewhere around maybe 500 Hz or so (depending on what I find works best), and then the subs, which are on the floor, coming in somewhere south of 200 Hz.  At that point, the "speaker" would be acting roughly like a floor-to-ceiling line.  The potential benefits of this configuration would be possible reduction of output in the low mid-range and upper bass for seats nearer to the speakers and reduced floor and ceiling interactions overall.  The speakers as they are hold the horizontal pattern quite well down to almost 500 Hz, so ideally I could continue to shadow the near seats in the lower frequencies too.  I really like this concept "on paper" but I haven't done a detailed analysis yet.

 

I like your idea, and it wouldn't be hard to implement in your case. I would be interested in hearing your subjective impressions of the pseudo line.

Share this post


Link to post
Share on other sites

So after reading this thread over the past year and amazed and the technical depth and extreme attention to detail paid to the tuning of this system and going "man I really want to hear this!", I flew and went to check out this system. 

And boy what an amazing system to listen to! My mind was blown as I was amazed by one thing after the other. All the work put into getting the tonal balance of this speaker correct really paid off big time. The whole system just sounds really "correct", and the more I listen to it the more I'm amazed by it. I brought my Reference Mini's with me as a comparison, and there was a very obvious difference in sound quality. I thought my speakers sounded really great, but it sound noticeably "off" when compared to this system. The speakers had a fantastic amount of detail, and the transients are awesome! It felt like I'm listening to a pair of really good headphones (and few people realize how hard and impressive it is to achieve this), but I also get the enveloping sound that makes speaker listening so pleasurable. It's the best of both worlds. 

What's even more impressive is the bass. I don't think I've heard bass so tight and full sounding in a room, which is clearly due to the complex integration efforts of multiple subs and individual EQ's to get such flat bass over a large number of seats. The clarity and tightness is seriously impressive. Again, just like a headphone, and that is actually something I've never heard before from a subwoofer. It is straight up the best sounding bass I've heard in a room. Now when you also get the whole body physical sensation from bass,  addictive is an understatement. 

One thing that is unforgettable and blew my mind is how great the speakers sound in the kitchen! I don't think SME has ever mentioned this, but it was indeed one of his goals. It was remarkable hearing a correct tonal balance with almost no treble roll off in a different room! I still can't believe this is achievable. It must be the combination of controlled directivity speakers and properly placed diffusers pulled this amazing magic trick of a feat. 

I've heard a lot of amazing home theaters, but this is the first time I heard imaging from surrounds. It was trippy to be able to pinpoint the location of the sound going across the rear stage. I really wish we watched an action movie and be able to so accurately track the position of the sound effects. This is even more impressive as I seem to clearly have less ability to hear imaging compared to other people. Speaking about imaging, the speakers reproduced phase manipulated music tracks far more accurately than anything I've heard so far. It must be the room treatments that are preserving the phase accuracy of the speakers. It was like "oh this is where it is supposed to sound!"

I was also exposed to the dark secrets of the time domain in room correction. That was a revelation to me to be exposed to so much more information and tools to analyze room acoustics. Now it makes sense why and how the room is mucking up the sound. It's all in the time domain! Now I am able to correlate measurements and subjective judgment of how good (or bad) the room sounds. I have so much to dig and play around with now. Measurements really can tell you about how good something sounds if you look at the right things and how to interpret it properly. 

Thank you SME and his wife for being such amazingly gracious hosts. That was one hell of a weekend! 

Oh, and did I make it clear enough that your system sounds good? :D

  • Like 2
  • Thanks 1

Share this post


Link to post
Share on other sites

Reading this (above) makes me happy.

SME builds, measures, makes adjustments, attention to details that surely can not make any difference - but it does. Then he - SME - describes the amazing sound, well, what does he know that all the others don't..

And the proof is in hearing and experiencing yourself. As lowerFE did. When you focus on the parts that are important for sound quality, and fix it, you actually get results that matters.

  • Like 1

Share this post


Link to post
Share on other sites
23 hours ago, lowerFE said:

Oh, and did I make it clear enough that your system sounds good? :D

Thank you for your kind words.  This process has been enormously challenging and time consuming, but results (so far) are immensely rewarding.  I never imagined sound of this level of quality was possible, in this space or otherwise.

I remember when I first moved into this house in December 2012 and set my system in the living room for the first time.  The sound was so disappointing.  My original plan was to try to finish part of the basement for a dedicated room, as soon as I could afford it.  However, the basement options were full of ugly compromises.  In one area, I would have 14 feet of width but would have to deal with ceiling obstructions with only 6 feet overhead clearance.  The other part offered a full 7 feet of headroom but only 10 feet of available width.  I had to come to terms with the fact that the basement options would be suboptimal regardless and decided to try to make the best of the living room space instead.

Today, I could claim to have a set up that approaches "world class"  performance, all while leaving the living room largely functional, albeit with lots of weird looking panels and diffusers.  Thankfully, my wife has been very accommodating.  Her skepticism toward acoustic treatments melted away once she heard the difference.  She also happens to be quite the bass lover, lucky me!

...

Some day I need to update the first posts on this thread to describe my "current" configuration.  Right before @lowerFE's visit, I migrated my speaker DSP configs to use FIR filters almost entirely.  I also modified the crossover to 850 Hz LR8 (acoustic).  The FIR filters are much cleaner and more precise than the mess of biquads I was having to maintain.  The horn/woofer crossover is also linear phase, which I opted for not so much for sound quality improvement but to eliminate group delay that confounded my tonal balance calibration method using short FDWs.  The result provided a significant improvement, albeit not as dramatic as some changes in the past.  Still, it was worthwhile enough for me to demo with the FIR filters, despite the fact that the bass still needed work.

So @lowerFE was able to hear the speakers sounding as good as they ever have, but the bass was not as good as I think it could have been.  In fact, I ended up making substantial changes on Saturday night, between his visits.  On Saturday, the main/sub XO was linear phase, and I ended up redoing everything to minimum phase XOs and less aggressive shaping to reduce pre-ringing.  That was kind of a hard lesson for me, which is that pre-ringing really does bad things to bass transient response and tactile sensation.  The problem was most obvious to me when listening to the Danley fireworks.  I could actually perceive the pressurization before the bang happened.  Even with those changes, some pre-ringing persisted and is present in my current config.  I don't know how perceptually important that is though.

Since @lowerFE left I've EQed down the 70-100 Hz range a bit, as it was subjectively too strong, but the bigger change was to move my bass boost from being centered at 70 Hz to being centered at 155 Hz instead.  I decided to try to better mimic the floor gain from a "typical floor standing speaker".  I had tried bass shelves at higher frequencies like that before, but it seemed to work a lot better this time.  The extra mid bass really brought more punch and overall loudness to the table.  Now I'm trying to figure out how to reduce pre-ringing further while maintaining smooth frequency response, keeping excess group delay in check, ensuring coherent summing across multiple channels, and doing all of this at every seat location.  It's a remarkably complicated problem, and while I have powerful DSP to attack it with, it's not at all obvious how best to apply this capability.  I also have a problem of a rattling window pane (at around 60 Hz, unfortunately), so I am trying to reduce the bass build-up in that corner to keep it from rattling as much.  I intend to eventually try to optimize using an automated algorithm, but automation is useless without a precisely defined objective.

And in the long run, I expect I won't be able to get the results I want with the equipment I have.  I still intend to replace the MBMs I have.  The open question is *where* I'm going to put the new MBMs.   I can put some of them behind the sofa like the old ones.  I can also put some of them on top of the subs, between the subs and left/right mains.  (The "pseudo-line" approach.)  And I can put some up on the shelve above the TV, adjacent to the center channel.  The locations behind the sofa are starting to fall out of favor with me because it's difficult to avoid pre-ringing problems.  In fact, I can't really avoid pre-ringing in the dining room and kitchen areas when using behind-the-sofa MBMs without using multiple switchable DSP configs, which I'd like to avoid.  So I'm curious if I can get away with MBMs on the front stage only.  I think the approach has potential, given how the center channel measures.  That is something I will investigate in due time.

Some time, I might start a thread about bass phase response / group delay.  It seems to be a substantially neglected issue with regard to system optimization and may have a strong bearing on tactile response performance.  While it seems counter-intuitive that minimum phase crossovers may (often) be superior for mains/sub crossovers, minimum phase systems actually appear to have the properties we want most.  We want as much energy as possible to arrive at the start of the impulse.  Too much positive excess group delay, and energy does not arrive until too late to contribute to perceived impact.  (Post-temporal masking effect.)  But any pre-ringing has the effect of shifting the perceptual reference point, the "start of the impulse", to a place where there's very little energy at all.  (Pre-temporal masking is very weak.)  So what achieves these goals?  For a particular magnitude response, the minimum phase response maximizes the amount of energy in the initial impulse.  I suspect that this is what's needed for the best tactile "kick".

  • Like 1

Share this post


Link to post
Share on other sites
9 hours ago, SME said:

Some time, I might start a thread about bass phase response / group delay.  It seems to be a substantially neglected issue with regard to system optimization and may have a strong bearing on tactile response performance. 

Pre-Subscribed...

JSS

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×