And you also need proper, preferable, applicable room acoustic treatments, I believe (just for your ref. here).
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
What is your favorite house curve
- Thread starter Jerry Sobel
- Start date
And you also need proper, preferable, applicable room acoustic treatments, I believe (just for your ref. here).
Yeah but regardless, the questions remains: 1) why in a poorly treated room and without equalization, the monitors sound so dark compared to a reasonably decent pair of hi-fi speakers, and 2) why couldn't I achieve with REW and EqualizerAPO a result at least comparable to what I got effortlessly with AM-1?
Regarding the second question in particular, Sure my inexperience may have played a role here, but I find it very odd that a software app can automarically balance the tonality of the speakers in the room so well (even more odd considering that the room is poorly treated). Because it means that there must be a method to infer the tonal balance of the speakers sounding in the room, and most of all, to infer what should be in that room the eq to apply that would return the best perceptually balanced tonality....
Have you measured the difference between your best effort and what MA-1 came up with?
I feel essentially same as you; I believe we should not over-trust advanced DSP and measurement-tuning software which includes/consists of many black-box-type (at least for myself) internal procedures/processings.
Just for one example case, as for the establishment of 1 msec to 0.1 msec precision time alignment over all the SP drivers, I would like to establish my own primitive but straightforward reliable reproducible fully-validated measurement/tuning method(s); if you would be interested, please refer to my following posts on my project thread.
- Precision measurement and adjustment of time alignment for speaker (SP) units: Part-1_ Precision pulse wave matching method: #493
- Precision measurement and adjustment of time alignment for speaker (SP) units: Part-2_ Energy peak matching method: #494
- Precision measurement and adjustment of time alignment for speaker (SP) units: Part-3_ Precision single sine wave matching method in 0.1 msec accuracy: #504, #507
- Measurement of transient characteristics of Yamaha 30 cm woofer JA-3058 in sealed cabinet and Yamaha active sub-woofer YST-SW1000: #495, #497, #503, #507
As you can find here and here, @zergxia carefully traced my methods in his audio system using the test tone signals I prepared, and he found/shared that EqualizerAPO does not always give him correct/right tuning information in terms of time alignment over the SP drivers.
Another example case would be Fq response measurement method(s); I usually do not like the popular quick sine sweep method and the psychoacoustic smoothing thereof. I have been sticking to rather primitive but very much reliable "cumulative (recorded) white noise averaging method" for Fq response measurements in digital level, analog line level, amplifier's SP-out high level, and actual room sound using microphone. Again if you would be interested, please refer to my post here, and my summary post here.
Hello again, @onununo,
I frequently compared the two methods, i.e. "rapid sine sweep" versus "cumulative (recorded) white noise averaging", at my rather large listening/living room (the multichannel full audio system) and at my rather small office upstairs (desktop DSP audio system); I found that the resulting Fq response curves obtained by the two methods are/were not always similar/identical with each other. And, the difference are/were always more significant at the small office upstairs which has very little room acoustic treatments.
You may easily guess that the "cumulative (recorded) white noise averaging" would better "simulate" our usual music listening situation since the SP drivers receive/sing "white noise", not the sharp/narrow-Fq sine sweep.
I would like to suggest/recommend you, therefore, to compare the Fq response measurements (at your listening position, of course) using the two methods, and you would assess which of the two methods would better fit for your subjective listening sensation.
At least for myself, I always prefer "cumulative (recorded) white noise averaging" giving better fit with my subjective listening sensation/feeling.
I frequently compared the two methods, i.e. "rapid sine sweep" versus "cumulative (recorded) white noise averaging", at my rather large listening/living room (the multichannel full audio system) and at my rather small office upstairs (desktop DSP audio system); I found that the resulting Fq response curves obtained by the two methods are/were not always similar/identical with each other. And, the difference are/were always more significant at the small office upstairs which has very little room acoustic treatments.
You may easily guess that the "cumulative (recorded) white noise averaging" would better "simulate" our usual music listening situation since the SP drivers receive/sing "white noise", not the sharp/narrow-Fq sine sweep.
I would like to suggest/recommend you, therefore, to compare the Fq response measurements (at your listening position, of course) using the two methods, and you would assess which of the two methods would better fit for your subjective listening sensation.
At least for myself, I always prefer "cumulative (recorded) white noise averaging" giving better fit with my subjective listening sensation/feeling.
Good point. No, but when I'll get back home after my vacations, I plan to try mimic with REW the FR I get from MA-1 and see if I can get somewhere. If I don't remember wrong, I tried already to do it once, but the results were not as good. It's true however, that Equalizer APO is not a minimal phase equalizer, so in order to avoid introducing phase differences from one channel to the other, I apply the same equalization to both channels, thereby equalizing just the "vector sum" at my precise listening position. I guess MA-1 does a better equalisation using the DSPs inside the subs.
Nevertheless, even if with REW+ EQ APO I'll finally succeed in getting the speakers to sound similar to what MA-1 comes up with, the question of not being able to determine "scientifically" what the correct "house curve" should be (something that AM-1 appears to know well, instead), remains...
My bet is it doesn't know this either.
It might however know your speakers extremely well. Something that is usually an unknown in generic house curve fitting approaches.
Hi @dualazmak,
I already noticed yesterday, as I was re-reading the whole thread, that you made a lot of work. Tried to follow what you made, but couldn't understand it. Being lazy on vacation, I quickly jumped to the conclusion that you made a measurement of your (plenty!) speakers' time of fly using white noise, as opposed to the common method using swept frequency with which to reconstruct the impulse response. Am I correct? Not sure why there should be a difference among the two methods however to behonest. I wouldn't take for granted that white noise better simulates the reality of music reproduction from speakers as opposed to swept frequency. Not sure either if what you mean for pulses are actually tone bursts instead, but again, quite likely I didn't pay the necessary effort to understand what you have done. I promise I'll try reading it through better.
No, you got me wrong. I am not making any such point. I'm just observing that there must be a well established ("scientific") method for equalizing any speaker in any room with the correct tonal balance (provided of course that we have all the necessary speaker data and we make the necessary measurements of the in-room sound, of course).
By "correct tonal balance" I mean the same tonal balance (i.e perceived amplitude equilibrium across the entire audio band) that would be perceived if that speaker were operating in anechoic environment. Despite of course some worse timbric performance, depending on how well or bad the room is treated. But the tonal balance should not be a matter of fiddling with the tone controls to find the setting that you like most!
Yes, this is what I am more keen to believe. It must know the speakers extremely well, and know how to deal with the 7 measurements it requires you to perform at precise spatial displacements. Specifically, it must know very well how to deal with that beast called phase.
OK, sorry for my misunderstandings on this point.
I believe/assume, however, your thought of "there must be a well established ("scientific") method for equalizing any speaker in any room with the correct tonal balance" would be unrealistic in the real world of our individual room; I assume we can never completely reproduce "anechoic" specification of a speaker in our "echoic" room environments...
It's just a matter of reproducing the overall tonal balance, not the "complete anechoic specification". Just the tonal balance, regardless of jaggyness of the FR due to the interaction of the direct field with the reflected field in the room. The right overall slope of the FR, i.e. the right perceptual amount of bass, mid and high frequencies relative to each other, should not be so hard to estimate!
OK, I understand your point; we possibly may somewhat simulate (manually or automatically) the best tone balance by EQ/XO/gain controls, I hope. In this context, I feel/assume subjective listening/tuning (like you are currently experiencing, right?) would surpass objective Fq response measurements.
The 0.1 msec precision time alignment over all the SP drivers, however, would be another critical factor in multichannel multi-SP-driver audio setup, but this factor maybe not so critical in "one box active studio monitor type" speakers which already so manufactured/established in quasi-completely time aligned.
Last edited:
By equalizing speakers subjectively, we're always prone to fool ourselves. And we'll never have a proof that our settings are right. It's like pretending to be able to calibrate the colorimetry of a display to, say, +/- 3 deltaE without a colorimeter and without any reference whatsoever that we can rely upon.
Regarding time alignment of multispeaker systems, that's not something I'd start discussing because it is not the main point here.
Sorry to be a little bit out of the scope of this thread and out of the thoughts of @onununo, but please allow me posting this.
As for the difference of popular "rapid sine sweep" (with much smoothing) and "cumulative (recorded) white noise averaging" (without typical smoothing but FFT size as smoothing factor), my recent post here would be of your reference, I assume.
As for the difference of popular "rapid sine sweep" (with much smoothing) and "cumulative (recorded) white noise averaging" (without typical smoothing but FFT size as smoothing factor), my recent post here would be of your reference, I assume.
OK, understood what is your major concern right now.
Generally speaking, however, not only tone balance (total Fq response) but also precision time alignment together greatly "matters" the total sound quality including darkness/brightness (your terminology), as well as transient characteristics, mutual harmonic distortions (especially around overlapped XO Fq), 3D sound imaging, stereo perspectives (or virtual disappearance of speakers), etc. You need to optimize (best tune) "the both" towards your targeting quasi-perfect SP (and room) tuning, I believe.
Yes, true, but as you already noticed, in my case I'm dealing with a 3-way system, whose speakers are already time aligned in factory, plus a pair of subs. I am sitting precisely at the same distance from all the speakers, subs included. In addition, I have matched the phases at crossover frequency (subs vs main speakers). So this puts an end to the story in my case.
Last edited:
I've done some experimenting lately, EQing my listening position to the Harman curve, with it's bumped up bass. Generally I like that, but after enough listening, especially to classical music, I reduced the bump up in the bass. The overall slope in the mids and highs still sounds good to me.
Yes, I understand it seems (sounds) fine.
May I ask what is/was your method for 0.1 msec precision time alignment measurement/tuning over all the SP drivers, and also how you could validate the accuracy/precision of the method?
By "validation" I mean, if you would intentionally delay (using DSP) your woofer in 55.5 msec against sub-woofer, it should be precisely measured as 55.5 msec delay in 0.1 msec precision with your measurement method.
You kindly wrote "I'm dealing with a 3-way system, whose speakers are already time aligned in factory"; have you double check it by your validated method?
Generally speaking, especially in passive XO-ed SP system, big woofer driver may possibly delay in 0.1 msec order against midrange driver because of the large difference in "inertial mass" (mass of moving parts) between the two crossover-ed drivers even if they receive sound signal exactly the same timing. (On the other hand, "such physical delay" can be usually ignored between midrange to tweeter as well as tweeter to supertweeter).
In my case, like your setup, my main SP is 3-way (well manufactured famous YAMAHA NS-1000), but I could find and adjusted 0.3 msec delay of 30 cm woofer against midrange-squawker(SQ) plus tweeter (ref. here).
Please note that in YAMAHA's original/intact passive XO design, they (YAMAHA) had no way to adjust this 0.1 msec order delay; as I made (DIY modified) the SP system fully active, i.e. each of the SP drivers is now dedicatedly driven by each amplifier in DSP multichannel setup, now I can/could establish 0.1 msec precision time alignment using group delay functions in upstream by using the DSP software.
I have also L&R large-heavy subwoofer YAMAHA YST-SW1000 (again like your setup) which was found to be in delay in 16.0 msec against the main SP (ref. here and here).
Consequently, for complete time alignment establishment, midrange-SQ+tweeter+super-tweeter has been delayed 16.3 msec, woofer has been delayed 16.0 msec, so that they have been perfectly time aligned with sub-woofer at my listening position.
Just for your kind attention, my time alignment measurement/tuning methods have been fully "validated" in 0.1 msec precision level.
Last edited:
I thought I was clear enough: I aligned the speakers by putting their acoustic centres at the same distance from the centre of my head in the listening position. Alignment can be done with delays in DSPs and positioning the speakers at any precision you want, even microseconds. However let me say that 0.1ms is just 3.4 cm in space, so it's nice to have but a little overkill, considering that your head will hopefully move a little, when you're on the sofa listening to your favourite music or enjoying your favourite movie.
There are plenty of ways to do that. Having had the chance to look a bit deeper in how you made it, yes, this is one. I guess another one could be using REW to play a swept sinusoidal tone covering the exact band of the speaker under test, then see the corresponding impulse transformate, from which getting to the delay information is a breeze (albeit I give you a point: I would not be able to tell the accuracy due to my fading memories in math). In such case of course the swept tone should be played with the option to first playing a high pitch tone from the tweeter, to set the reference "0" for timing, just as you have made. Such option is already included in REW.
My speakers (Neumann KH310A) are active type studio monitors. Each driver has its own amp inside, and I am assuming that the manufacturer has already included time delays in the amps to time-align them. Ok, given that these speakers have no DSP inside, there is indeed a chanche that this assumption is wrong. However in any case there's not much I can do to improve the time alignment if I discovered that it is bad, considering that there's a single common analog input on the box and for sure I'm not going to open the speakers and pull out the 3 individual inputs. Yes, I could check the time alignment of the 3 drivers, for instance with your method, but for the reason I just explained, I don't see much point in even bothering.
Yes, I see your point regarding the need to check the time alignment of the subs with the main speakers by measurements, so I'll make some when I'll be back from the vacations. However the only thing I could do to adjust the relative timing from subs to mains is moving the subs back and forth, because the XO circuit from subs to mains is inside the subs, and there's no option for time delay/group delay compensation, but only a variable phase compensation. I guess everything is done by the automated MA-1 calibration software, considering that it acts on the DSP parameters inside the subs (so setting a delay/performing group delay compensation, is possible), and that such software is also specifically designed for dealing with that model of main speakers connected. I agree however that the use of MA-1 calibration suite limits the freedom of what can be done manually. Neverthess, it beats any manual EQ I managed to apply so far.
So If I understand well what you mean for validated, you can demonstrate that you can reach time accuracy of 0.1 ms in your measurements. Fair well...
EDIT: some typos / some rewording
Last edited:
- Joined
- Jun 10, 2018
- Messages
- 6,244
- Likes
- 9,377
Choice of a house curve is important because if one is going to equalize there has to be a base to equalize to. One must also consider how much. Is 1 dB the minimum deviation allowed, or 2 dB? How much smoothing should be used? I use 2 dB and psychoacoustic smoothing. For many around here that isn't enough. Some of the equalization coefficients proposed to increase a loudspeaker's preference score to the maximum leave me wondering if there is such a thing as too much. There are many adjustments of under 1 dB with a Q as high as 12. Could that cause ringing and distortion?
Similar threads
