I'm not trying to go against Drs. Toole and Olive at all. The problem is, how do you do those w/o an anechoic chamber and a sophisticated computer program? The best approximation I or any of us not in the business can do is the gated polar response--and it is useful as they show serious resonance, diffraction, on axis smoothness/flatness and I can't imagine how they wouldn't hint at power smoothness. I'm certainly not saying I'm utterly correct here, but it's the best I can figure how to do and it seems to correlate with what I hear and Dr Toole writes. For the low end, DIYers are essentially useless at measuring barring near field. I don't know how useful any near field measurement is. Dr. Toole is certainly not a fan, but I can't say I really understand why. It would be nice if the engineers in the business insisted on the correlated measurements being posted in the product literature. Sure would make buying speakers simpler for the informed. The uninformed might learn as well in a few years.
Dave, any input you can give on how the home DIY guy can do the measurements best correlated to listener preference would be greatly appreciated.
As far as which sounds most correct (to me😀) in my polar graphs--the Mackie followed by the JBL(Tony, I think you may have been reading the wrong title as my labeling isn't the best). I'd say the KRK is possibly the worst, but maybe the popular B2031P.😱 They are lacking in different ways. Too many ways to screw up a speaker. The cotton certainly helps it out, but improving the Rokit's issues wouldn't be as simple. As the volume gets loud, the Mackie are the nicest. My other Mackie doesn't have the huge resonance above 10 kHz. Funny thing is, on the recording boards people seem to have different preferences out of the monitors I have listed and they utterly have no correlation with measured performance. Typically they hate the Mackie monitors and really like the JBL.😕 Reasoning typically goes something like: Mackie makes good stage monitors and mixing boards, JBL makes great speakers for all purposes. Side by side they are more similar than any of the others to my ears. They defend their choices just as passionately as the listeners do. With all the lack of standards, it's no wonder sound reproduction still has so many problems. Maybe that's what makes it so fun and addictive?
Dan
This is a subject that interests me a lot: how best to measure frequency response. There have been some great papers on how we perceive frequency balance in a live room by Kates, Lipshitz and Vanderkoy, Salmi, Bech and others. The general consensus in this group is that we perceive frequency balance as if we used a time window on the arriving sound. The window is short enough at high frequencies that it is just the direct sound. By mid frequencies it is long enough to include the floor bounce. By low frequencies it is generally long enough to include all room effects, or the steady state sound.
If that is how we hear, it suggests that we should measure in a similar way: focus on the direct sound from 1kHz and up, the steady state room curve from 200 down, and a little of both for mid frequencies.
If you have Holm you can gate the response for direct sound above 1k. Holm with a long window (or no window and 1/6th Octave smoothing) will show the steady state room response for LF. If you don't have Holm but have some kind of RTA, you can take a curve from the listening position and know what you need to know about low frequencies. You can then move towards the speaker, being careful to stay on the listening axis, and converge towards a spot midway between mid and tweeter to get a response curve that is primarily the direct sound. Between those two measurements you know a lot about the response of the system.
If I listen first and measure second, I find those two curves always show pretty well what I like or don't like about the system.
The only problem with steady state room curves is they lean much more towards the power response of the system and the acoustics of the room, even though your ears are pretty good at hearing through to the direct sound.
David S.
there never will be the speaker that pleases all unless some type of world wide brain and emotional zombie controller is invented and released at the same time..
instead of forcing your house sound or what you feel is important on too others, try to understand that the next guy may find that it stinks and go the exact opposite direction..
or make large posters of your response graphs and put them were your speakers should have gone as it seems that it's gravitating less towards the music and more towards dyno queens..
just my .02$
instead of forcing your house sound or what you feel is important on too others, try to understand that the next guy may find that it stinks and go the exact opposite direction..
or make large posters of your response graphs and put them were your speakers should have gone as it seems that it's gravitating less towards the music and more towards dyno queens..
just my .02$
Last edited:
Thanks Dave! That's exactly what I do already--it's all in my blog. I just wish it was possible to produce the exact curves of Toole and Olive. Of course I'd also like to get better resolution all the way around, but have no plans on building and calibrating an anechoic chamber any time soon.
There's a bit more to perception in the HF than just the direct sound--the reflections can increase apparent source width, spaciousness, envelopment, intelligibility and detail according to Toole's work which I'm sure I'm paraphrasing a bit. I gotta agree with him there as well as it's what I've heard with the speakers shown. I believe these HF things are all sort of on a curve but in most rooms you really won't reach a true problem with a well designed speaker. I always listen before I measure as well. With the recent JBL I could have pretty much drawn the curves, but not with the KRK Rokit 6. I just knew it reproduced voices in a strange fashion that I could apply several adjectives to. I just wasn't sure what to make out of what I was hearing, but it definitely sounded a lot different from the others.
Thanks again,
Dan
There's a bit more to perception in the HF than just the direct sound--the reflections can increase apparent source width, spaciousness, envelopment, intelligibility and detail according to Toole's work which I'm sure I'm paraphrasing a bit. I gotta agree with him there as well as it's what I've heard with the speakers shown. I believe these HF things are all sort of on a curve but in most rooms you really won't reach a true problem with a well designed speaker. I always listen before I measure as well. With the recent JBL I could have pretty much drawn the curves, but not with the KRK Rokit 6. I just knew it reproduced voices in a strange fashion that I could apply several adjectives to. I just wasn't sure what to make out of what I was hearing, but it definitely sounded a lot different from the others.
Thanks again,
Dan
Last edited:
My question to you guys is does a speaker sound more real simply because it has flat on-axis response? To me, the answer is no. A speaker with a flat response can still sound dead and unintersting, whereas one with an irregular response can be fun to listen to.
There's got to be more to how our brains process musical information than simply the tone of an instrument. In my experience, if it has to sound real, it has to reproduce information in the time domain accurately. In other words, it sounds worse if the timing is not right compared to if you have an irregular frequency response.
There's got to be more to how our brains process musical information than simply the tone of an instrument. In my experience, if it has to sound real, it has to reproduce information in the time domain accurately. In other words, it sounds worse if the timing is not right compared to if you have an irregular frequency response.
They're not fun because of the non-flat response. They have qualities which are not
immediately obvious which make them fun. What would these qualities be? I'm not sure
yet, but certainly better time domain behavior could be one of them.
It is like the old argument of valves v/s SS. A good single-ended triode definitely sounds
good and has its followers. If THD was the only criteria and everyone thought the same,
these things would not exist. For that matter, my DeLite amp with the light bulb also
sounds similar, although not quite in the same league. But even with its higher distortion
it is quite enjoyable... it makes music. It makes music unlike some higher power, ultra-low
distortion amps.
Low distortion is an important aim, so is flat frequency response. But let's not get complacent
by saying that is the single most important goal. If it were so, nobody would like full range drivers.
immediately obvious which make them fun. What would these qualities be? I'm not sure
yet, but certainly better time domain behavior could be one of them.
It is like the old argument of valves v/s SS. A good single-ended triode definitely sounds
good and has its followers. If THD was the only criteria and everyone thought the same,
these things would not exist. For that matter, my DeLite amp with the light bulb also
sounds similar, although not quite in the same league. But even with its higher distortion
it is quite enjoyable... it makes music. It makes music unlike some higher power, ultra-low
distortion amps.
Low distortion is an important aim, so is flat frequency response. But let's not get complacent
by saying that is the single most important goal. If it were so, nobody would like full range drivers.
Ra7, there's no way for anyone to reply what you are saying. Your asking what the qualities would be of a non-flat measuring speakers that make them enjoyable for you to listen to. You'd probably the best person to answer your question. I'd bet what I think would be far different from what you'd think, but maybe not. Perhaps you could support your statements with some graphs and listening impressions?
Dan
Dan
Yeah... I know what you mean... I'm not quite sure I can provide graphs and such.
I'm not that advanced in my learning as yet.
All I'm saying is that we need to explore some of the good things coming out of full
range drivers, one of which is point source nature and good coherence. I've heard
some good controlled directivity systems which still don't sound as coherent as a
single driver.
Thanks for being patient with me.
I'm not that advanced in my learning as yet.
All I'm saying is that we need to explore some of the good things coming out of full
range drivers, one of which is point source nature and good coherence. I've heard
some good controlled directivity systems which still don't sound as coherent as a
single driver.
Thanks for being patient with me.
Ra7, If you have little time distortion, automatically your speakers measure flat (but the inverse is not true).
It's because the drivers are (supposed) minimum phase devices. Then, any irregularity of the amplitude curve has it's equivalent in the phase curve. Full range drivers could escape this law only if not behaving as band pass (e.g linear from 15Hz to 25000 Hz). But you know, as some of them sound really nicely while having by nature these phase shifts, maybe time distortion is not the absolute key, just one of them.
try to read the big JK's site, starting by this one.😉
try to read the big JK's site, starting by this one.😉
There is no one measurement more important than the other, the answer is within the mix and the balance you get from such.
speaker-room interface:
http://www.diyaudio.com/forums/multi-way/121385-loudspeakers-room-system.html
Again, Olive studied the various measureable factors of a large group of loudspeakers and found the following factors and weighting predicted rank order in a listening test.
Narrow band anechoic axial response smoothness weighted 31.5%
Narrow band room response smoothness (20.5%)
Wideband in-room response smoothness i.e. flatness (17.5%)
Low bass extension (30.5%)
Measure them, weight them all together and the numerical score correlates highly with controlled listening test scores. (Correlation 0.86)
They are all some form of frequency response. Measurements of phase response, polar response (at least directly), or distortion were not needed to rank order speakers. Therefore those are low level or nonexistent factors.
David S.
Technical note: you can always make your statistics sausage machine produce numbers like that. Might be true or might be "common causation" or other pitfalls of proof-by-correlation. The question is whether it can be justified by others and by related test methods.
Elsewhere, the Harmon team suggest good bass trumps all other factors in their kind of testing.
I am greatly impressed by their work... up to a point. As a methods-guy, I've been struggling to understand how good judgments of speaker quality can be made from over-processed pop female vocalist recordings, pink sound, yet way down the list of good test sounds, traditional classical music groups.
Elsewhere, the Harmon team suggest good bass trumps all other factors in their kind of testing.
I am greatly impressed by their work... up to a point. As a methods-guy, I've been struggling to understand how good judgments of speaker quality can be made from over-processed pop female vocalist recordings, pink sound, yet way down the list of good test sounds, traditional classical music groups.
Last edited:
I would be pretty excited if I came up with a numerical ranking scheme with high correlation. I certainly haven't seen anybody come up with any other plausible factor analysis.Factor analysis - Wikipedia, the free encyclopedia
Its one thing to say "I believe it's about phase response" or whatever. Its another to try prove any association numerically. Without the numerical association it is only oppinion. This is pretty understood in most other fields, such as medicine, but often ignored in audio.
David
- Status
- Not open for further replies.