Conceptually one can reach a house curve with any loudspeaker, so in that sense one might believe that the loudspeaker response does not matter. But Toole did a study where the speaker was optimized for its anechoic response and then put in a room and a speaker which did not have a good anechoic response was equalized in the same room to match the anechoic response speaker. there was a strong preference for the anechoic optimized speaker. To me this makes perfect sense because the direct sound - the first arrival sound, before the room even enters into the picture, must be correct for proper imaging etc. I.e. the anechoic response must be correct. One cannot correct a bad speaker design by simply EQing to a "preferred" room curve. This of course is only true above the modal region - below that the exact opposite is true, the speakers response is all but irrelevant and only the room response matters.
I did a quick test of this using rePhase to implement FIR convolution. Used a 12" coaxial with a 1k 48LR.
On very select passages I could hear slight differences. Close miked guitars were the easiest for me to pick up on. Also noticed differences in the lower octaves but preferred the non corrected version for whatever reason. At 24LR I couldn't reliably tell a difference.
Conclusion. I'm not going out of my way to implement linear phase.
@gedlee
Your speakers seem to be excellent in all areas that really matter. One question I have would be about using a slightly larger diaphragm for those low crossover points. The new DE550 looks like it would be a good compromise for larger rooms.
I've seen many speaker FR's both on axis and polars. Many of the good ones I've seen that have very good on-axis FR's also have reasonably smooth off-axis polars as well.
Here is but one example from Mr. Atkinson's vast measurement library. You may take issue with his methods now that you've developed your own polar measurement system and that's okay. I visited your site but found I needed to download the polar S.W. and some database files in order to view the polars you have. I gave up at that point.
Dynaudio Excite on axis, left and polars, right.
Attachments
Last edited:
Thanks.
If these were speakers for Pro use then I might agree, but for home use the smaller diaphragms are fine. The larger diaphragms do not go as high up as the smaller ones and thus tend to need another driver for the upper most frequencies - I don't like that idea.
That's too bad, because I think that Atkinson's polars are flawed and that my technique is the best out there. Your loss I guess.
The example that you show is actually pretty bad, but with plots like that it is not easy to see.
Last edited:
I would agree that linear phase is on the edge of what is audible - not very important for the most part, except in some extreme cases. Way down on the list of priorities.
True...I could never get a super tweeter to sound right. I'd rather live with coherent output to 13k instead of a tweeter.
@Lojzek As noted in the stereophile article: “…Many loudspeakers are claimed by their manufacturers' marketing departments to be time-coherent. There are also a number of speakers that have sloped front baffles, implying that they are time-coherent. However, its step response immediately gives you an indication of whether or not a loudspeaker is time-coherent (on the chosen measurement axis). And almost all loudspeakers are not...”
@PeterCharles re: “…then a linear phase response would also result in a perfectly flat group delay in the time domain.” Exactly!
@ErnieM – would be cool to see the step response from your experiment…
The OP asked, “But what is meant by 'accurate' here?” Maybe folks have a different idea what accurate means, but for me, it means the music waveform arriving at my ears is as close as possible to the original music waveform that is stored on the source media.
As one can see from my previous post, to hear a “perceived” flat frequency response at the listening position, there is a so called house curve. Over a span of +40 years, that curve has remained the same or very similar, regardless of acoustic environment (we are talking about small room acoustics here). Personally, I have listened to the B&K curve in dozens of environments, from semi anechoic to very lively. The end result is still the same, perceptually flat frequency balance at my ears. The caveat is within reasonable limits of what would be deemed a critical listening environment.
On the REW, Audiolense, Acourate forums, I have corresponded with recording engineers, mastering engineers, audio enthusiasts, with a wide range of speakers systems and listening environments. Most would agree with the +40 years of research as to what sounds perceptually flat at the listening position (which is all I care about as I don’t listen in anechoic chamber).
How that is achieved can be a matter of debate. The most sophisticated software DSP systems produce correction filters that equalize the direct sound at higher frequencies, but applies a frequency dependent window that widens as frequency decreases. There is also some excess phase correction applied.
It certainly is possible to eq a system to the so called house curve with less than stellar results, just have a look at the step response (and group delay: http://en.wikipedia.org/wiki/Group_delay_and_phase_delay ) and one can see why this is the case. I have run into this problem myself several times until I started investigating the timing aspect of sound reproduction.
Here is an example of measuring the frequency response of my right speaker at the listening positon, same speaker system but the red curve is with 3 way digital XO, time alignment, and eq and the blue curve is with a 3 way passive XO:
Both in the same ballpark from a B&K house curve perspective, especially above 300 Hz.
Let’s look at the step responses, again red is digital time alignment and blue passive XO of the same speaker system:
Huge measured, and certainly to my ears, audible difference. The red step response shows all frequencies arriving at the same time at the listening position, which is clearly not the case with the passive XO in blue.
Which is my point. Accurately reproducing a stored media’s music waveform at the listener’s ears requires us to take into account both the frequency and time (i.e. step) response.
In music this is called timbre: http://en.wikipedia.org/wiki/Timbre From a timing perspective, note the section on http://en.wikipedia.org/wiki/Timbre#Envelope
How can we expect to accurately reproduce music waveforms at our ears if the speaker systems timing (i.e. step response) messes up the waveform?
@PeterCharles re: “…then a linear phase response would also result in a perfectly flat group delay in the time domain.” Exactly!
@ErnieM – would be cool to see the step response from your experiment…
The OP asked, “But what is meant by 'accurate' here?” Maybe folks have a different idea what accurate means, but for me, it means the music waveform arriving at my ears is as close as possible to the original music waveform that is stored on the source media.
As one can see from my previous post, to hear a “perceived” flat frequency response at the listening position, there is a so called house curve. Over a span of +40 years, that curve has remained the same or very similar, regardless of acoustic environment (we are talking about small room acoustics here). Personally, I have listened to the B&K curve in dozens of environments, from semi anechoic to very lively. The end result is still the same, perceptually flat frequency balance at my ears. The caveat is within reasonable limits of what would be deemed a critical listening environment.
On the REW, Audiolense, Acourate forums, I have corresponded with recording engineers, mastering engineers, audio enthusiasts, with a wide range of speakers systems and listening environments. Most would agree with the +40 years of research as to what sounds perceptually flat at the listening position (which is all I care about as I don’t listen in anechoic chamber).
How that is achieved can be a matter of debate. The most sophisticated software DSP systems produce correction filters that equalize the direct sound at higher frequencies, but applies a frequency dependent window that widens as frequency decreases. There is also some excess phase correction applied.
It certainly is possible to eq a system to the so called house curve with less than stellar results, just have a look at the step response (and group delay: http://en.wikipedia.org/wiki/Group_delay_and_phase_delay ) and one can see why this is the case. I have run into this problem myself several times until I started investigating the timing aspect of sound reproduction.
Here is an example of measuring the frequency response of my right speaker at the listening positon, same speaker system but the red curve is with 3 way digital XO, time alignment, and eq and the blue curve is with a 3 way passive XO:
Both in the same ballpark from a B&K house curve perspective, especially above 300 Hz.
Let’s look at the step responses, again red is digital time alignment and blue passive XO of the same speaker system:
Huge measured, and certainly to my ears, audible difference. The red step response shows all frequencies arriving at the same time at the listening position, which is clearly not the case with the passive XO in blue.
Which is my point. Accurately reproducing a stored media’s music waveform at the listener’s ears requires us to take into account both the frequency and time (i.e. step) response.
In music this is called timbre: http://en.wikipedia.org/wiki/Timbre From a timing perspective, note the section on http://en.wikipedia.org/wiki/Timbre#Envelope
How can we expect to accurately reproduce music waveforms at our ears if the speaker systems timing (i.e. step response) messes up the waveform?
Could this be one of the reasons that some people prefer speakers with a single point source
In Stereophile article it is also noted :
"Of the 350 or so loudspeakers I have measured, there is no correlation
between whether or not they are time-coherent and whether or not they
are recommended by a Stereophile reviewer. However, I feel that if other
factors have been optimized—on-axis response, off-axis dispersion,
absence of resonance-related problems, and good linearity—like a little
bit of chicken soup, time coherence (hence minimal acoustic phase error)
cannot hurt."
Earl has made a very good point reminding us what Toole discovered about
people preferring the speaker optimized for its anechoic response over the
one not but equalized to match the anechoic response speaker.
This is the optimal way of making an accurate speaker.
I remember an interview with Dr. Joe D'Appolito where he was asked what he
thought about time alligning the drivers. His answer was people are trying to
solve problems that don't exist. An excellent achivement of loudspeaker design
is to have a smooth response of direct sound.
You can read about it in German here:
Interview HOBBY HIFI mit "D'Appolito" - Visaton Diskussionsforum
Last edited:
Completely agree with above. The point is, the distortion of the frequency response, i.e., linear distortion, is so gross that minor timing errors don't really matter. Besides, our ears are not sensitive to the type of distortion creating by timing errors, at least not until it becomes gross. For example, having separate drive units may result in slightly different time arrivals, but the ear cannot pick it up. Whereas, the ear can pick up the change in frequency response. Now, if your drivers are over a feet or two feet apart, then it might start to become important. Remember that subwoofers need not be time aligned. Only their phase near the crossover matters.
Yes of course the differences are clearly audible - you went from a complete disaster to something reasonable. But is no evidence that phase integrity is a requirement. The fact is that there is still that no valid subjective tests that have shown that absolute phase integrity is audible. The "waveform" does not have to be highly accurate as if our ears were microphones, because they are not. Ears are a very course sensor that has its own set of distortions which mask a great many "waveform errors". Toole and Olive discount phase completely. I don't discount phase integrity completely, but I don't give it high importance either.
I would suggest that we be careful here. As my wife and I showed in a large blind subjective study, group delay becomes more audible at higher SPLs. So when we say that timing errors are not audible at all we have to clarify at what SPL are we talking about. This aspect gets completely lost in ALL of these phase discussions. Speakers that "lose-it" at higher SPLs (like most of them) will most likely never have audible phase anomalies - the phase issues will be masked by the other issues. But a speaker that can play extreme SPLs without any more error than they have at low levels just might have audible phase anomalies.
As some of this discussion on phase may have come out of one of my earlier comments, I thought I had better qualify my statement. I was talking about phase matching at the crossover frequency and a bit either side. Not what some class as the holy grail of linear phase.
Whilst it may be possible to make a crossover with poor phase matching at the crossover point (lets say 45 deg out of phase) sum flat on axis, it probably won't produce a very good result.
Of course the above may fall into the category of "broken" and should be picked up if off axis response is examined, but the reason I mention it, is because I think a lot of people starting out with crossover design only worry about getting a flat on axis response, and if they have that, think that all is good. In my opinion (based on my limited experience) that's not the case
Tony.
Whilst it may be possible to make a crossover with poor phase matching at the crossover point (lets say 45 deg out of phase) sum flat on axis, it probably won't produce a very good result.
Of course the above may fall into the category of "broken" and should be picked up if off axis response is examined, but the reason I mention it, is because I think a lot of people starting out with crossover design only worry about getting a flat on axis response, and if they have that, think that all is good. In my opinion (based on my limited experience) that's not the case
Tony.
Perception of harmonics can be impacted by the relative phase of the harmonics.. eg Schroeder and Mehgart "Auditory Masking
in the perception of speech" show that reversing the phase of even one harmonic component is audible. He could even "produce little melodies by sheer phase manipulations." Their theory is that the signal may have small "silent" intervals where even a weak signal can be perceived i.e. unmasked. This is an important distinction that is not predicted by critical band masking theory, in fact, are actually inconsistent with critical band masking theory
From "Hearing, Its Psychology and Physiology" by the
Acoustical Society of America
""...The qualitative character of the audible change produced by adding this harmonic was different at the various sensation levels of the fundamental. At low levels the harmonic was usually heard as a separate tone. In the middle region [50 - 80 dBSPL] it was heard as a sharpening or brightening of the timbre of the tone, whereas at high levels the changes were so complex and so dependent upon differences of phase that any generalization about their character would be misleading."
So I think what may be happening is the increased distortion at higher levels creates a signal in which group delay may become more audible.
Dave
I saw the Toole presentation that I believe you are referring to.
He had two units, both with a slightly messy directivity index curve. One was equalized to have superb axial response and slightly messy power response. The other was equalized to have superb power response (with the same general downhill trend of the first). This meant that neither system was perfect but the first had the best direct response and the second the best power response. His blind testing panel gave a slight preference to the unit with better direct response.
I am certainly of the camp that thinks that the direct response should be as smooth and flat as possible. I have also read many studies that conclude that we hear in a time windowed way, giving preference to the early sound (as I have stated here many times before). At the same time I have been doing some testing on the topic and not getting the result I would have expected.
I have created a test system with fairly directional forward firing elements and separate rear firing elements. Using this in the near field I am able to adjust independently the direct and reflected sound balances. I fully expected that there would be a difference between, say, having a sensible room curve with normal directivity and dishing out the treble of the direct sound while replacing it with later arriving rear sound.
Well, after several months of trying I can't really say that I am detecting much difference between room curves arrived at with different balances between direct and reflecting spectra. Certainly a hole in the direct response can be well filled by the right amount of later arriving energy. When taken to extremes then the spatial impression is different but the steady state curve seems to define frequency balance, at least in rooms of domestic size.
What does this mean?
First and foremost, we can't get too pedantic about speaker directivity and polar patterns. If an on-axis dip can be corrected with and off-axis peak, then polar smoothness is not essential. Also, energy spectrum of particular reflections don't seem to matter, as long as total response is correct, again implying that polar performance is a loose descriptor
Constant directivity (at least above a certain frequency) can work but it isn't the only answer.
Now the direct to reflecting energy ratio is quite audible, especially if it can be directly adjusted. I found that equal energy between direct and total reflected energy is about as far as I would want to go on classical music and too far for pop (the spaciousness is fun but you end up preferring the precision that goes out the window).
Flat anechoic response with falling power response is a safe recipe for a good sounding speaker. My tests (and Lipshitz and Vanderkooy's) show that the total power response needs to roll downhill and that dips in power are generally benign if the axial response is flat.
At the same time the implication is there that a nonflat axial response, allied with the right power response, can also give a well balanced sound.
Since the direct sound is independent of the room (and the customer's room is unknown), our safest approach is to design speakers with flat axial response and falling power response. It will work in a greater percentage of rooms.
It just might not be the only solution.
David S.
Optimizing for the direct on axis response while caasing larger response errors off axis creates a less accurate loudspeaker, in a small reverberant room. Large open room where first reflection delay is high, not so much.
IME many hobbyist designers do exactly this.
It would be interesting to see what Dr. Toole's on axis equalization did to the first reflection response and the t30 response. Without that, it's hard to know if he traded one problem for another
Dave
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.