davewantsmoore Posted June 24, 2023 Posted June 24, 2023 (edited) 52 minutes ago, March Audio said: if not its the most badly behaved tweeter I have ever seen. I'm guessing you corrected the response from an in room measurement, which is the wrong thing to do. https://audioxpress.com/assets/upload/images/1/20210624153600_Figure3-SBAcoustics-Satori-AT60NC-4-AMT-Tweeter.png That is what voicecoil found. It's ok. Even using an in-room measurement could be "good enough for jazz" if it was windowed and from an appropriate distance(s).... <shrug> Edited June 24, 2023 by davewantsmoore
Grizaudio Posted June 24, 2023 Posted June 24, 2023 (edited) No point discussing further until full range response measurements are provided. Should be interesting. Edited June 24, 2023 by Grizaudio
stereo coffee Posted June 24, 2023 Posted June 24, 2023 3 minutes ago, Grizaudio said: No point discussing further until full range response measurements are provided. Should be interesting. Or music with full range response is heard
March Audio Posted June 24, 2023 Posted June 24, 2023 2 minutes ago, stereo coffee said: Or music with full range response is heard That won't help at all.
Grizaudio Posted June 24, 2023 Posted June 24, 2023 (edited) 1 hour ago, March Audio said: guessing you corrected the response from an in room measurement, which is the wrong thing to do. Stepping away from Andy’s discussion for a sec… Both yourself and Dave have stated this….. If you listen from a static seated position what are the main concerns correcting for a single listening position? I.e to Harman target… or preference even. I understand that you are likely correcting for things you shouldn’t, that could potentially be better addressed by speaker location, speaker choice, Roon treatment etc….. But Full range correction is the goal of many room correction solutions, I.e Dirac live…. And would likely provide an improved response if done correctly. ….. can you please validate this comment, so I can get on the same page… thanks Steve Edited June 24, 2023 by Grizaudio 1
March Audio Posted June 24, 2023 Posted June 24, 2023 (edited) 2 hours ago, Grizaudio said: Stepping away from Andy’s discussion for a sec… Both yourself and Dave have stated this….. If you listen from a static seated position what are the main concerns correcting for a single listening position? I.e to Harman target… or preference even. I understand that you are likely correcting for things you shouldn’t, that could potentially be better addressed by speaker location, speaker choice, Roon treatment etc….. But Full range correction is the goal of many room correction solutions, I.e Dirac live…. And would likely provide an improved response if done correctly. ….. can you please validate this comment, so I can get on the same page… thanks Steve What you hear at the listening position is a combination of the direct on axis sound from the speaker and the off axis sound reflected from the room boundaries. So whilst the on axis sound is of primary importance, the off axis sound is also extremely important. The on axis *anechoicly measured* sound should be, as you might expect, flat and smooth. However, speakers emit sound in all directions, but not uniformly. The character of this dispersion has a big influence on the overall sound. The off axis sound, whilst it will almost certainly have a falling response towards high frequencies, should also be smooth. The overall sound power should be smooth. Speakers that have directivity problems will not have a smooth off axis response and generally will not be preferred. Good speakers with good on axis and good off axis responses will create a falling frequency response when measured in room. It's been demonstrated that this is what is subjectively preferred. The degree of slope will be individual to the specific speaker and specific room acoustics. Never eq a room response to flat, it will just sound horribly bright. So, to get to your point about EQ systems. EQ can correct for fundamental on axis frequency response issues. *BUT* what it can't do is correct for any directivity issues. So, EQ won't fix bad Speakers or their effect at the listening position. Basically, only use listening position measurements to adjust for issues below about 200Hz, ie to knock down room modes. Above 200Hz only perform subtle adjustments of the slope of roll off over a wide frequency range. Never try to correct small dips and peaks as it's likely to be the measurement position issue. To correct for speaker fundamental frequency response issue you need to use free field measurements. Ie in an anechoic chamber, in free space such as an open field an a good 6m off the ground, or by using Klippel system which removes the effect of the room from the measurements. Edited June 24, 2023 by March Audio 1
Keith_W Posted June 24, 2023 Posted June 24, 2023 3 hours ago, Grizaudio said: Stepping away from Andy’s discussion for a sec… Both yourself and Dave have stated this….. If you listen from a static seated position what are the main concerns correcting for a single listening position? I.e to Harman target… or preference even. I understand that you are likely correcting for things you shouldn’t, that could potentially be better addressed by speaker location, speaker choice, Roon treatment etc….. But Full range correction is the goal of many room correction solutions, I.e Dirac live…. And would likely provide an improved response if done correctly. ….. can you please validate this comment, so I can get on the same page… thanks Steve I agree that you should not do an overall speaker correction from the listening position. This is because, as wavelengths get shorter, what the mic measures is a model of what you actually hear. A mic which is fixed in one position with a mic tripod is not the same as two ears on a head which turns, moves around, lies down, etc. Remember that a mic is almost omnidirectional, whereas what arrives at your eardrums has been modified by your pinna, ear canal, reflections from your shoulders, masked by the shape of your face, size of your nose, etc. So for high frequencies, what you measure is not actually what you hear. I have done this experiment myself, I have tried: averaging the response over a large area and doing a full range correction (this should more closely approximate a pair of ears). I have also tried doing a quasi-anechoic measurement using beamforming (as described by Krol in his paper). In every case, doing a full range correction results in the system becoming too "top endy" despite applying a target curve. Note that there is a strong possibility of user error here, given that I am hardly a DSP guru. At the moment, my approach is to start off with the "perfect speaker", i.e. nearfield driver measurements with windowing and then driver linearization only. Get the crossover sorted so that everything looks perfect. And then LEAVE IT ALONE and only do correction below Schroder. 1
andyr Posted June 24, 2023 Author Posted June 24, 2023 2 minutes ago, Keith_W said: I agree that you should not do an overall speaker correction from the listening position. This is because, as wavelengths get shorter, what the mic measures is a model of what you actually hear. A mic which is fixed in one position with a mic tripod is not the same as two ears on a head which turns, moves around, lies down, etc. Remember, Keith ... even though 'sweet spot' optimisation is what is required for my system (as it's not a 'home theatre' setup) Con took several mic measurements either side of the 'head' position - then averaged them for the feed into his Convolution program.
March Audio Posted June 24, 2023 Posted June 24, 2023 5 hours ago, davewantsmoore said: A planar in a round waveguide, without a baffle.... have a look in the speaker thread (it might explain a lot). Yes that round baffle is significantly impacting the tweeter response. I guestimated the dimensions and put into Virtuixcad.
Keith_W Posted June 24, 2023 Posted June 24, 2023 1 hour ago, andyr said: Remember, Keith ... even though 'sweet spot' optimisation is what is required for my system (as it's not a 'home theatre' setup) Con took several mic measurements either side of the 'head' position - then averaged them for the feed into his Convolution program. Yeah Andy, there are different approaches. We discussed some of those approaches today at the 3SB GTG. What works for me and my system may not work for you, given that the configuration is so different. I think that every person who does DSP has to decide for themselves what works best for them. A lot of it comes down to preference as well, for e.g. I like the clean sound that I get from extremely steep X-over filters. Aris prefers the smooth sound of shallower filters.
davewantsmoore Posted June 24, 2023 Posted June 24, 2023 4 hours ago, Grizaudio said: If you listen from a static seated position what are the main concerns correcting for a single listening position? I.e to Harman target… or preference even. The only thing I would care to clarify or add to Alans post on this (which I totally agree with) ... is that it depends somewhat on whether you are "designing a speaker" (ie. correcting raw drivers) ... or if you are correcting (with a full range correction) some already designed speaker, where the directivity and sound power issues that Alan mentioned have already been taken care of by the designer..... ie. when you ask "should I be doing this thing or not" .... it depends somewhat on what you are trying to do / achieve. If it is the former (get some drivers, make a speaker).... then no-way. As Alan says, this is not going to give a good result..... you need to account for directivity and diffraction, and how these things interact for the multiple drivers, from the get-go.... and how these things combine to inform an overall target curve for the speaker (measured at whatever axis) to get the correct constant or smoothly falling sound power. You can't measure the drivers in the listening position, with a few averages.... you can't measure the drivers up close and correct them, etc. etc... and expect a great (or perhaps even very good) result. If it's the later (got a finished speaker, want to correct it in the room) .... then it isn't a total no-no, obviously things like Dirac Live, and other "correction systems" work like this. But the same issues mentioned by Alan apply. The problem, which I have mentioned many many times - so greatly affects the sound... is the choice of target curve for these systems. It depends a lot on your speaker. When you are measuring the at the LP, the what you're getting is the result of the directivity/power.... and it's quite hard to know what target curve will not wreck the speaker (or "help" the speaker if it not so well designed). Systems like Dirac Live, which take a bunch of measurements and process the data, try avoid correcting tiny (or big) things which should not be so corrected, in my extensive fuzzing of them, do a pretty good job .... and I think it is not sacrilege to let them run full range, but the choice of target will have a profound effect. The paradox with all of this "correction stuff", is that whatever measurement you take (whether for "speaker design", or "in room correction", you can correct to be flat (or whatever target) and it will look "perfect".... but the problem is that the measurement is not representative of anything relevant. The second paradox with all of this is that many speakers are junk, and straightening them out, even with an all the problems mentioned, and an ill conceived target curve..... could easily sound "improved" The third and similar paradox, is that choice of any target for EQ, will have a profound impact on sound ... and hence plays very strongly to peoples preference (for and against) .... a bit like saying that sensibly seasoned high quality food is one thing.... but the super-seasoned vs unseasoned, is bound to invoke "preferences" more in people (strong like or dis). 3
davewantsmoore Posted June 24, 2023 Posted June 24, 2023 4 minutes ago, Keith_W said: Yeah Andy, there are different approaches. We discussed some of those approaches today at the 3SB GTG. What works for me and my system may not work for you, given that the configuration is so different. I think that every person who does DSP has to decide for themselves what works best for them. A lot of it comes down to preference as well, for e.g. I like the clean sound that I get from extremely steep X-over filters. Aris prefers the smooth sound of shallower filters. There are many ways to skin a cat.... We can decide what we like... and nobody can really argue with that easily. The "approach" is basically irrelevant, if you look at it through the lens of "is the result a well designed speaker"..... ie. does it actually work for you. ... but what constitutes a well designed speaker (ie. the definition of "work") .... isn't as "flexible". 1
March Audio Posted June 24, 2023 Posted June 24, 2023 (edited) 24 minutes ago, davewantsmoore said: The only thing I would care to clarify or add to Alans post on this (which I totally agree with) ... is that it depends somewhat on whether you are "designing a speaker" (ie. correcting raw drivers) ... or if you are correcting (with a full range correction) some already designed speaker, where the directivity and sound power issues that Alan mentioned have already been taken care of by the designer..... ie. when you ask "should I be doing this thing or not" .... it depends somewhat on what you are trying to do / achieve. If I undestand correctly this is new design, so the drivers (as installed in the frame) would need to measured in free field conditions. A correction applied (with crossover) and the woofer and tweeter levels matched. You will then need to measure in free field again and and make sure the crossovers sum correctly. You could then do a listening position measurement and further correct for room modes. What none of the above does of course is tell you what the directionality of the speaker is doing. It could be good, bad or indifferent, which is going to have a massive impact on the sound. To find this out you need to perform spinorama measurements. Its quite possible to do this in your garden with a turntable. Basically, designing speakers is no trivial task. You cant just throw a bunch of drivers together, apply an arbitrary 12dB/octave crossover in followed by an overall correction based on listening position measurements. Edited June 24, 2023 by March Audio 2
Grizaudio Posted June 24, 2023 Posted June 24, 2023 @March Audio thanks for taking the time to reply. Understanding broadly the importance of uniform directivity, direct and indirect sound is one reason I have the speakers I do. My question was really relating to correcting a well sorted speaker (I.e. In my case a speaker with decent uniform directivity) to target, not speaker design perse. But I understand the interplay you have described, and the importance of considering these factors. 23 minutes ago, davewantsmoore said: The only thing I would care to clarify or add to Alans post on this (which I totally agree with) ... is that it depends somewhat on whether you are "designing a speaker" (ie. correcting raw drivers) ... or if you are correcting (with a full range correction) some already designed speaker, where the directivity and sound power issues that Alan mentioned have already been taken care of by the designer..... ie. when you ask "should I be doing this thing or not" .... it depends somewhat on what you are trying to do / achieve. My question was asking about the later. 28 minutes ago, davewantsmoore said: If it's the later (got a finished speaker, want to correct it in the room) .... then it isn't a total no-no, obviously things like Dirac Live, and other "correction systems" work like this. But the same issues mentioned by Alan apply. The problem, which I have mentioned many many times - so greatly affects the sound... is the choice of target curve for these systems. It depends a lot on your speaker. When you are measuring the at the LP, the what you're getting is the result of the directivity/power.... and it's quite hard to know what target curve will not wreck the speaker (or "help" the speaker if it not so well designed). Systems like Dirac Live, which take a bunch of measurements and process the data, try avoid correcting tiny (or big) things which should not be so corrected, in my extensive fuzzing of them, do a pretty good job .... and I think it is not sacrilege to let them run full range, but the choice of target will have a profound effect. This paragraph is really the answer to my question. 50 minutes ago, davewantsmoore said: The problem, which I have mentioned many many times - so greatly affects the sound... is the choice of target curve for these systems. It depends a lot on your speaker. When you are measuring the at the LP, the what you're getting is the result of the directivity/power.... and it's quite hard to know what target curve will not wreck the speaker (or "help" the speaker if it not so well designed). Does this need to be complicated Dave? Obviously target curve choice is important, and listener preferences will most certainly influence preferred targets. It has been my tendency to select a target curve which best mirrors the natural response of the speaker above say Schroder, and then EQ bass to suit preference. In my case Harman curve reflects this preference well. Other than considering preference, and natural speaker response in room - what is the better method for target curve selection in your opinion?
March Audio Posted June 24, 2023 Posted June 24, 2023 (edited) 1 hour ago, Grizaudio said: Other than considering preference, and natural speaker response in room - what is the better method for target curve selection in your opinion? I think the point is that given a well designed speaker that has good on and off axis responses and sound power response, the in room measurement will naturally follow close to the right curve. The Harman curve is only a guide and the final result will depend on the room acoustics and absorption. Good speakers don't need much if any EQ, apart from dealing with the inevitable room modes. Edited June 24, 2023 by March Audio 3
davewantsmoore Posted June 24, 2023 Posted June 24, 2023 12 hours ago, March Audio said: If I undestand correctly this is new design Yes. I was just clarifying more generally for others, who might be confused be "it's always a wrong thing". 12 hours ago, March Audio said: so the drivers (as installed in the frame) would need to measured in free field conditions. Sure. So the coverage pattern of the woofers can be matched to the coverage pattern of the tweeter, etc... and the overall target / power response can be understood, etc. etc. etc. (as you know, of course). It's actually a really "complex" speaker. Dipole woofer(s) that are all playing up to the XO point, matching to a monopole ... and diffraction. 12 hours ago, March Audio said: What none of the above does of course is tell you what the directionality of the speaker is doing. It could be good, bad or indifferent, which is going to have a massive impact on the sound. To find this out you need to perform spinorama measurements. ... and this is where many are making the biggest "mistake", that I see .... looking at the directivity once the speaker is "finished"..... as opposed to looking at the directivity of the individual drivers on baffle as almost the first step in the design ... and then changing the shape/size/layout of the baffle, driver EQ, and crossover choice... to optimise said directivity (as you know, of course). 12 hours ago, March Audio said: Its quite possible to do this in your garden with a turntable. ... or even, at a big stretch, in a room with a v. short window. (problematic, and not recommended, of course .... but can give you the general) 3
davewantsmoore Posted June 24, 2023 Posted June 24, 2023 12 hours ago, Grizaudio said: Does this need to be complicated Dave? It is not simple. For example. Let's say you take "the harman speaker curve".... and you force the LP measurement of your speaker to that curve. The harman curve slopes the way it does (rather than being totally flat) because it is compensating for the measurement that you will get at the LP .... but this depends on the directivity, distance, room, etc. etc. It might be "about right" over some of the curve.... but it's hard to say. The paradoxical(?) thing is that correcting to a target like this, is a "low Q" + "significant amplitude" correction.... it will be highly audible. It's going to make you say "oh wow, I can hear things in the recording that I never heard before... I love it" .... or say "oh yuck, the speaker sounds like it's EQed" .... or something. Is it "always wrong/bad" ... "should you do it?" ... "will you like it?"... It's hard to talk in absolutes. I use Dirac Live... but I always look at before and after from multiple angles, both in room, and outside.... I don't want it to change very much, or at least want to understand what's going on, rather than picking anything based on "sounds good". 12 hours ago, Grizaudio said: listener preferences will most certainly influence preferred targets. I think this is a dangerous game... and I'd lean much more towards buy a good designed speaker and don't "fix" it. 12 hours ago, Grizaudio said: It has been my tendency to select a target curve which best mirrors the natural response of the speaker above say Schroder, and then EQ bass to suit preference. Sure... but there even issues with just fitting a target curve over the top of your LP measurement. Only before and after measurements can really show "what happened". 12 hours ago, Grizaudio said: In my case Harman curve reflects this preference well. All this "no no" being said.... if this target doesn't sound too different from the uncorrected speaker (without "bass eq") then it's probably ok. 12 hours ago, Grizaudio said: what is the better method for target curve selection in your opinion? As I said, it isn't simple. The directivity needs to be considered. You need be very sure that the "in room" data is not obscuring important things (it may very well be). At the end of the day, "preference" is what most people will use.... but big changes in sound are probably problematic. 2
Grizaudio Posted June 25, 2023 Posted June 25, 2023 (edited) 13 hours ago, March Audio said: I think the point is that given a well designed speaker that has good on and off axis responses and sound power response, the in room measurement will naturally follow close to the right curve. The Harman curve is only a guide and the final result will depend on the room acoustics and absorption. Good speakers don't need much if any EQ, apart from dealing with the inevitable room modes. Thanks Allan, Applying this discussion to my application..... My Speaker: JBL M2 My Room: 3x3.5m, standard height ceilings, carpet floor. LP is a 2.1m nearfield equilateral triangle. Acoustics: 50mm Broadband absorption on front, rear, and top left right walls. I'm moving soon, and plan to treat ceilings, side wall first reflection points with diffusion once that happens. Below are two quick measurements for my system, taken at approximately seated head height @ LP. I would estimate this could be approximately 10degrees off vertical axis... The non eq'd response displays a natural tilt, and slow roll off from 10khz to 20khz... I think a portion of the 10-20kh roll off is due to the of axis measurement, if I measure on axis at 1m, the compression driver if far more linear up top. These measurements were taken on a whim, just to see what the system was doing in room, and to test generate inverse convolution filters. My next goal is to take multiple measurements @ head position, and create a vector average and re do my inverse convolution filters up to Schroder. After listening with and without full range target curve correction, I feel most of the benefit is below 300hz. To provide some context, this is my LEFT channel response, no EQ. Speaker is located in a corner, hence the bass loading. Right channel, speaker is located in a corner, next to an empty'ish built in robe. Edited June 25, 2023 by Grizaudio
Grizaudio Posted June 25, 2023 Posted June 25, 2023 (edited) 1 hour ago, davewantsmoore said: It is not simple. For example. Let's say you take "the harman speaker curve".... and you force the LP measurement of your speaker to that curve. The harman curve slopes the way it does (rather than being totally flat) because it is compensating for the measurement that you will get at the LP .... but this depends on the directivity, distance, room, etc. etc. It might be "about right" over some of the curve.... but it's hard to say. Thanks Dave, I think I'm clear now. For instance, here is 'a' version of the Harman target curve (there are a few kicking around), overlayed with my left and right responses - probably measured about 10 degrees off axis @ seated head height. No averaging on responses here - just a single measurement. Clearly where and how you measure is going to heavily influence the curve response measured, and hence what is corrected. So you could be correcting for totally the wrong things. I might take a few measurements directly on axis at seated position and see how the response changes relative to target, just for interest sake. I think a separate thread on target curve correction could be very interesting. Edited June 25, 2023 by Grizaudio
March Audio Posted June 25, 2023 Posted June 25, 2023 11 minutes ago, Grizaudio said: Thanks Allan, Applying this discussion to my application..... My Speaker: JBL M2 My Room: 3x3.5m, standard height ceilings, carpet floor. LP is a 2.1m nearfield equilateral triangle. Acoustics: 50mm Broadband absorption on front, rear, and top left right walls. I'm moving soon, and plan to treat ceilings, side wall first reflection points with diffusion once that happens. Below are two quick measurements for my system, taken at approximately seated head height @ LP. I would estimate this could be approximately 10degrees of vertical axis... The non eq'd response displays a natural tilt, and slow roll off from 10khz to 20khz... I think a portion of the 10-20kh roll off is due to the of axis measurement, if I measure on axis at 1m, the compression driver if far more linear up top. These measurements were taken on a whim, just to see what the system was doing in room, and to test generate inverse convolution filters. My next goal is to take multiple measurements @ head position, and create a vector average and re do my inverse convolution filters up to Schroder. After listening with and without full range target curve correction, I feel most of the benefit is below 300hz. To provide some context, this is my LEFT channel response, no EQ. Speaker is located in a corner, hence the bass loading. Right channel, speaker is located in a corner, next to an empty'ish built in robe. The M2 is a well designed speaker. Can you explain what you feel you need to change about the sound? The in room measurements are very close to what you would hope to see. However the full free field Klippel measurements do reveal an interesting characteristic. If you look at the sound power (red line) you will see it stays very flat as do the early reflections. Most speakers will show a sloping sound power. It's directivity is good, but does dip at crossover. Widening out 2 to 6kHz does make the sound power flatter in that range. Overall this will lead to a flatter in room response with less of a slope. Subjectively it may make the speaker sound a little bright. Also the tweeter response is very slightly elevated compared to the woofer which may reinforce this subjective observation. i have read subjective reports this is how the speaker sounds. Very good, just a tad bright. So what I would do is put a shelf filter in above 700Hz and bring the entire tweeter level down about 0.5dB to 1dB. This may not seem a lot but spread over that wide range it will be audible. What you don't want to do is use EQ to create a steeper in room slope. This will just mess up the good on axis response. Then just deal with the room modes bel8w 200 Hz. Don't over do that mode correction as it will lean out the sound. So the other thing that will affect the in room measured slope is the room acoustics. Can you post the RT60 measurement which will provide an indication of how absorptive the room is.
davewantsmoore Posted June 25, 2023 Posted June 25, 2023 1 hour ago, Grizaudio said: Clearly where and how you measure is going to heavily influence the curve response measured, and hence what is corrected. Yes. And also if you imagine a target curve the same-ish shape as the one you posted... but it was 2 dB higher between 200 and 1000 (just to pick some semi-random numbers) .... this would totally change the sound. Would that be right/good?.... You'd have to look at free field measurements, and in room measurements from multiple angles to really understand what was going on. You could just go with whatever you like most, but I think this is much more problematic than people give credit for. I find it to be a "trick", and that I always find that what "theory" says should sound better, actually does sound better (and if it doesn't, it will turn out of done something wrong). It isn't simple. 1 hour ago, Grizaudio said: I might take a few measurements directly on axis at seated position and see how the response changes relative to target, just for interest sake. I would highly recommend anyone going down this path to use something like Dirac Live.... where there is a system which intelligently processes multiple data points to form a correction. It isn't fool proof, of course, and I still think these systems are potentially quite problematic..... but it is much better than "manual mode". 1 hour ago, Grizaudio said: Diffusors Look cool.... In a small room, you're either sitting too close to them (at lower frequencies), or at higher frequencies the directivity is too high, and/or you don't want diffusion at those Hz anyways. 1
Grizaudio Posted June 25, 2023 Posted June 25, 2023 (edited) 4 hours ago, davewantsmoore said: I would highly recommend anyone going down this path to use something like Dirac Live.... where there is a system which intelligently processes multiple data points to form a correction. It isn't fool proof, of course, and I still think these systems are potentially quite problematic..... but it is much better than "manual mode". I haven't tried using Dirac Live multichannel software with Roon yet. My understanding is Dirac Live will present a Dirac Audio Processing endpoint within Roon, which in turn addresses the multichannel audio interface. Its something I would need to look into. I need to experiment with how this interferes with my execution of FIR crossovers in Roon. Or indeed whether these same crossovers can sit in the Dirac Live software as either IIR or FIR. Its something I have never looked into. My gut tells me, careful application of correction using inversion would provide similar outcomes, especially if targets are similar. 4 hours ago, davewantsmoore said: Look cool.... In a small room, you're either sitting too close to them (at lower frequencies), or at higher frequencies the directivity is too high, and/or you don't want diffusion at those Hz anyways. Yep... No diffusion atm. This is for my next space Dave. Thanks for exploring the topic with me. Much appreciated. Steve Edited June 25, 2023 by Grizaudio
Grizaudio Posted June 25, 2023 Posted June 25, 2023 4 hours ago, davewantsmoore said: And also if you imagine a target curve the same-ish shape as the one you posted... but it was 2 dB higher between 200 and 1000 (just to pick some semi-random numbers) .... this would totally change the sound. Would that be right/good?.... You'd have to look at free field measurements, and in room measurements from multiple angles to really understand what was going on. You could just go with whatever you like most, but I think this is much more problematic than people give credit for. I find it to be a "trick", and that I always find that what "theory" says should sound better, actually does sound better (and if it doesn't, it will turn out of done something wrong). I'll try correcting just up to Schroder - say 300hz.... I will also dig into Dirac Live multichannel a little to see how I manage both crossover and correction elegantly.
Grizaudio Posted June 25, 2023 Posted June 25, 2023 21 hours ago, March Audio said: The M2 is a well designed speaker. Can you explain what you feel you need to change about the sound? The in room measurements are very close to what you would hope to see. However the full free field Klippel measurements do reveal an interesting characteristic. If you look at the sound power (red line) you will see it stays very flat as do the early reflections. Most speakers will show a sloping sound power. It's directivity is good, but does dip at crossover. Widening out 2 to 6kHz does make the sound power flatter in that range. Overall this will lead to a flatter in room response with less of a slope. Subjectively it may make the speaker sound a little bright. Also the tweeter response is very slightly elevated compared to the woofer which may reinforce this subjective observation. i have read subjective reports this is how the speaker sounds. Very good, just a tad bright. So what I would do is put a shelf filter in above 700Hz and bring the entire tweeter level down about 0.5dB to 1dB. This may not seem a lot but spread over that wide range it will be audible. What you don't want to do is use EQ to create a steeper in room slope. This will just mess up the good on axis response. Then just deal with the room modes bel8w 200 Hz. Don't over do that mode correction as it will lean out the sound. So the other thing that will affect the in room measured slope is the room acoustics. Can you post the RT60 measurement which will provide an indication of how absorptive the room is. Hi Allan, I don't subscribe to the M2 as sounding bright at all.. I think this is a matter of taste, and depends highly on room and how loud you listen. My room isn't perfect. Its setup as a home office and listening space atm. This will change when I move home, and hopefully have a larger dedicated listening room. The room has mild treatment; 8 broadband absorbers front and rear, with small 300mm square 20mm absorbers at high level. First reflection points on side walls are unfortunately windows/cupboard doors. So they remain untreated. Treatment helped ALOT to combat audible ringing, and reflections. I usually listen with blinds down, to reduce interaction with glass. Room: Rear facing I quite enjoy the natural M2 response, as long as bass response is smooth and about 8ish dB higher overall. My listening is mostly done around 65-85db. I don't listen hyper loud. My original reason for measurement was just to see what the speaker was doing in room; so curiosity really. I didn't have any goal to apply correction at this stage. If anything I felt I could hear a lack of presence from 10khz (compared to my older active 3 way speaker, a vastly different design, dome mid, dome tweeter, 12inch woofer. Clearly directivity, amplitude differences (response extended flatter out to 20khz) and room interaction was all different. As highlighted earlier, if I measure the M2 smack bang on compression axis, the compression driver measures ruler flat, but the natural roll off above the 10khz-15khz presence hump is still there. Its there on the sound power charts above too. Just an FYI, centre of compression driver/waveguide is circa 1060mm AFFL, my ear @ seated position is the same - so the measurements I have taken are likely circa 5 degrees off vertical/horizontal axis. Over the last few days, I have been listening again only with the OEM crossovers engaged. It sounds amazing, I have no complaints, I think I'll re run the measurements, create a vector average and EQ using convolution inversion below Schroder only. All the measurements provided below and previously, was a single pass, so no averaging across my seated position. Something I still need to do. Uncorrected RT 60 right: Clarity Uncorrected RT60 Left: Clarity
Grizaudio Posted June 26, 2023 Posted June 26, 2023 (edited) In contrast this predicted RT60 and waterfall. Inverse convolution has been applied. I haven't measured to validate this... yet. Left Channel Right Waterfall is greatly improved Left Channel Edited June 26, 2023 by Grizaudio
Recommended Posts