I have read some but not not all (hence my questions)
Project at present is a jbl 4435 resemblance clone
two 15" drivers in each cabinet and buttcheecks sat on top
their is 1/2 meter between the 15" drivers horizontly then 1/2 meter between bass/mid and HF buttcheecks verticly,
one 15 upto about 100hz the other 15 including the same and upto around 1000hz both in the same cabinet space
trying to measure and implement them is just not happening
can not take them outside to measure due to weight
You should probably come back 5 feet or 1.5 meters on a line between the driver centers and your listening space. I.E., on the listening axis, but not too close. Aim for a more or less flat response there.
Then do measurements out at the listen position. Try several mic positions and average them, or uses the Moving Microphone Measurement. At that position you can then EQ to find the falling response that sounds good in the room.
Then do measurements out at the listen position. Try several mic positions and average them, or uses the Moving Microphone Measurement. At that position you can then EQ to find the falling response that sounds good in the room.
I have used a cheap Kenwood pseudo 1/3 octave analogue (digital interface) EQ. It is enough to dial in the response to within a fraction of a dB of where it sounds natural, after which I can incorporate the changes into the crossover and start over.
The differences from recording to recording are typically of a different nature, ie it should still sound right, just more or less bright if that makes sense.
If you can't get it right with some simple EQ it may indicate a speaker issue.
If your system is free of early reflections then the 1m response should be more or less the same at the listening position, except where 1m is too close to properly incorporate diffraction from the speakers, plus a small discrepancy at the very top end.
Head shadowing occurs for live sound sources as well. That in itself is not a source of error. The hrtf will transform the sound from one speaker in mono the same as it would for a live source at that location.
Its the effect of having two speakers reproduce the same sound at the same time that causes the hrtf to distort the tonal balance.
I investigated this myself a number of years back: what tonal error is introduced in stereo due to the HRTF for a central image (probably the worst case)?
Here it is attached. Either this gets "fixed in the mix" or speakers designs should modify their responses away from flat to try and strike a balance between correct tonal balance for the center image vs that for panned full left or right by applying "HRTFEQ". Given so much of the image clusters in the center, I think best bet is provide about 2/3 equalization.
I think this is one reason why so many speakers "flat" on axis sound harsh on a lot of recordings.
I hope people benefit from this information in their own designs!
Dave
Attachments
Thanks Dave. Good info. 
I have a thread about the phantom center image tonality with a link to some useful info.
But I don't think it matters about one speaker or two for overall tonal balance. The microphone does not hear the way we do. We have two "mics" separated by a head. That changes the tonal balance of whatever we hear, compared to a single point. At least above about 1200-1500 Hz, it does.
The sum from 1 speaker or two ends up about the same.
I have a thread about the phantom center image tonality with a link to some useful info.But I don't think it matters about one speaker or two for overall tonal balance. The microphone does not hear the way we do. We have two "mics" separated by a head. That changes the tonal balance of whatever we hear, compared to a single point. At least above about 1200-1500 Hz, it does.
The sum from 1 speaker or two ends up about the same.
Hi Pano. Very true that a mic will pick up a very different reverberant field than your ears will, so this is where its left to the craft of a good recording engineer to get the tonal balance right. But I think that doesn't invalidate this effect and I think the tonal balance question presents a different answer with one vs two speakers, especially for close mic'ed recordings
I certainly hear a different tonal balance with one vs two. It's a related issue, but not the same I think. Even with a single speaker what we hear is the sum of on axis and off axis ears. That summed response falls in the upper end. With two speakers things get even trickier. The upper midrange dip in the phantom center is one artifact. It's caused by comb filtering.
One might suppose that if music mastering suites had speakers that measured flat at the mastering position, then using speakers at home that mimic that flat response might give a good tonal balance. Alas, this seems not to be the case.
One might suppose that if music mastering suites had speakers that measured flat at the mastering position, then using speakers at home that mimic that flat response might give a good tonal balance. Alas, this seems not to be the case.
Two things are important in this regard, because there are two basic signals that the ear evaluates. The first is the direct field, i.e. that signal which reaches the ear and is evaluated as quickly as the ear can evaluate it. The 1 m anechoic response is the best indicator of this. The second is the reverberant field, which is closely related to the speakers power/polar response. This is what one gets with a steady state in-room measurement. They are both important and neither one can tell you the whole story.
Agree, but that said, the curve proposed by B&K decades ago, and by Olive recently, seem to do very well for most home listening. Sometimes simple is all you need.
Yes, exactly this. If you go back to post 31 http://www.diyaudio.com/forums/mult...at-vs-accurate-hi-fidelity-4.html#post4258668 One can see that not only the B&K curve, Toole, Olive, but EBU Tech 3276. Additionally, several forum members here and on other "room correction" forums have consistently found the same result - using this target will give the listener a perceptually flat, meaning all frequencies sound equal to the ear, response at the listening position.
As someone who has spent 10 years recording and mixing in a variety of studios/control rooms, there was at least one set of monitors eq'd to the B&K or EBU target, as they are effectively more or less the same. The reason being that recording and mixing using that "house curve" gave the best "translation" to a wide variety of speakers and listening environments, including car stereos. Also, from the engineers perspective, if I had to move from one studio to the next, having the same tonal response meant I did not have to start my mixes from scratch. Or worse yet playing your mix for the first time to the record company on a set of monitors that were way off not only made your mix sound like crap, but doubts creep in as to which set of monitors are the truth...
With respect to where to measure. As a consumer listening to music, at the listening position is the place to be. Further, our understanding of direct versus reverberant sound or psychoacoustics has increased and the use of frequency dependent windowing (FDW) where the window is large at low frequencies, but quickly narrows to the direct sound as frequency increases has greatly improved the analysis of the frequency response at the listening position to better reflect what our ears hear. Acoustic measurement software like REW, DRC, Acourate, Audiolense, HolmImpulse to name a few have FDW built in.
There is a lengthy but good discussion about FDW and psychoacoustic smoothing started by Bob Katz, mastering engineer, at: Feature Request: Frequency Dependent Windowing - Home Theater Forum and Systems - HomeTheaterShack.com Check out the first post and it's attachment to understand why FDW is important in measuring in-room frequency responses at the listening position.
for future reference.
the B&K curve is roughly this:
0db at 1khz
-1db at 2khz
-2db at 4 to 5khz
-3db at 6 to 8khz
-4db at 10khz
-5db at 20khz
There's also the curve recommended by mitchba and Bob Katz which is a bit more aggressive.
0 at 1khz
-2 at 2 to 3.5 khz
-3 at 4khz
-4 at 7khz
-5 at 10khz
-6 at 20 khz
there's also the JBL/ Harman/ Toole curve:
0 at 1khz
-1.5db at 2khz
-2db at 3khz
-3.5 at 4khz to 11khz. This is odd. from 4 to 11khz, its fairly flat....
-5 at 15khz
the B&K curve is roughly this:
0db at 1khz
-1db at 2khz
-2db at 4 to 5khz
-3db at 6 to 8khz
-4db at 10khz
-5db at 20khz
There's also the curve recommended by mitchba and Bob Katz which is a bit more aggressive.
0 at 1khz
-2 at 2 to 3.5 khz
-3 at 4khz
-4 at 7khz
-5 at 10khz
-6 at 20 khz
there's also the JBL/ Harman/ Toole curve:
0 at 1khz
-1.5db at 2khz
-2db at 3khz
-3.5 at 4khz to 11khz. This is odd. from 4 to 11khz, its fairly flat....
-5 at 15khz
Last edited:
Yes I very much agree. Many years ago (20?) I discovered the need to design in a small depression on axis where the tweeter starts to dominate on its low end due to the wider dispersion vs the mid or woofer. Its common practice now but it back then it wasn't and took allot of head scratching to figure out.
I seem to find that the smaller the room, the more weight the DI has on the tonal balance. Is that your experience as well?
Thanks! I traced the B&K curve a few years back and found this:
-0.5 at 20Hz
0.0 from 42-200Hz
-1.0 at 400Hz
-1.5 at 600Hz
-2.0 at 820
-2.5 at 1200 Hz
-3.0 at 1750 Hz
-3.5 at 2900 Hz
-4.0 at 4K
-4.5 at 5.3K
-5.0 at 7K
-5.5 at 10K
-6.0 at 15K
-6.5 at 20K
As long as that is the general trend, it works well. You can change the tonal balance by raising or lowering the hinge point.
Eventually I found a nice solution, using overdamped but high-enough order crossover filters: lots of overlap through xover so the excess tweeter outputs is offset by the lesser mid/woofer for nice off axis, but they transition out of band to their asymptotic slope and provide the needed power handling
Works gangbusters. George E. Short at North Creek fessed up that he came to the same design conclusion independently as well.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.