Thats not what you use in your paper, you use Coherence. And I know of no work that say that this or the MTF has been shown to correlate with perception. Your are simply assuming that a higher value of coherence is better. In room acoustics for instance a lower coherence would imply better spaciousness, i.e. lower coherence is "better". Unless you show the connection between what you maesure and subject and that it is stable, you are simply showing a bunch of numbers that could mean anything.
Yes, there is a version available for free on my home page. It can be used to plot wavelets from impulse response . It does not have modulation analysis though (the one I posted just now).
May I know what software you use to measure impulse responses ?
ASCII to WAV conversion is the least of difficulty. Does the file have a header ? Can you post a few lines from such a file, so I can make a conversion script ?
- Elias
John
Its an easy matter to convert a text file into a wav file with MathCAD. If you don't have that software I can do it for you because I do it all the time.
I found Elias plot interesting but unlike him I did not see anything that I thought was conclusive in the results. I did not see what significance a modulated wavelet has. There might be something there I am not sure, but it wasn't obvious to me.
The bass in a large room is always very good and yet its impulse response at LFs is very long. So why is a shorter LF impulse response in a small room a "good thing"? I'm just not convinced that these "metrics" are a valid surrogate for perception. I have not seen any tests that prove that they are.
Is it? Listening to a bass solo in a church is probably not a very pleasant experience.
Last edited:
This I wrote in diyaudio in August in another thread:
To repeat, I'm not using MTF in the traditional sense. MTF is another steady state signal analysis, not very usefull when music is being considered. In music the modulation is temporally compact, not continuous.
What you propably need to do is first analyse any music sample to find out the temporal aspects of the modulation it contains.
And what you say:
is generally incorrect and may apply only to interaural difference (similar to IACC).
Here the first hand question is how to reproduce the information from the recording in a small room. The information = modulation = music.
- Elias
Last edited:
I use the old Liberty IMP. I'll have to see what the format is. I believe it has a header and some info at the end about sampling rate and number of samples. That is easy enough to edit out. This I believe it just has a single column of data which is the sampled impulse. I'll that a look tomorrow (Sunday).
Hi Elias,
I completely agree with your graphs comparing impulse responses between dipoles and monopoles, as being the most representative of subjective perceptions, by many including me. For music with rapidly changing bass lines, indeed a dipole "sounds" as the one with the most definition, in any room big or small.
However, perhaps the only reason dipoles ( in passive systems)are not so common, is because of the limited bass extension needing equalization. From the views of John and the quoted paper , "Backman, J. Low-frequency polar pattern control for improved in room response, AES
paper 5867, 115th AES Convention, October, 2003", it seems a cardioid is the best midway compromise between a sealed and pure dipole (H-frame of flat panel) in terms of better bass extension than a dipole but better transient response than sealed.
It would be most interesting to know if you have used your excellent software to compare a sealed and cardioid bass unit, as your experiments are indeed the most representative of subjectve perception by many.
Thanks for your remarkable contributions and measurements !
I completely agree with your graphs comparing impulse responses between dipoles and monopoles, as being the most representative of subjective perceptions, by many including me. For music with rapidly changing bass lines, indeed a dipole "sounds" as the one with the most definition, in any room big or small.
However, perhaps the only reason dipoles ( in passive systems)are not so common, is because of the limited bass extension needing equalization. From the views of John and the quoted paper , "Backman, J. Low-frequency polar pattern control for improved in room response, AES
paper 5867, 115th AES Convention, October, 2003", it seems a cardioid is the best midway compromise between a sealed and pure dipole (H-frame of flat panel) in terms of better bass extension than a dipole but better transient response than sealed.
It would be most interesting to know if you have used your excellent software to compare a sealed and cardioid bass unit, as your experiments are indeed the most representative of subjectve perception by many.
Thanks for your remarkable contributions and measurements !
Soundaatma,
I am planing to do a comparison of monopole (sealed), dipole and cardioid in the same room assuming I can get my impulse files converted to the necessary wav format. I'll also have to learn how to run Elias' code and Octave.
Perhaps the better way to do this would be for me to just email my impulse files to Elias for processing? I still want to learn how to use Elias' code but I'm interested is seen the results. Still I have to set up the woofer systems which will take a little time. I'm a little busy at the moment.
I am planing to do a comparison of monopole (sealed), dipole and cardioid in the same room assuming I can get my impulse files converted to the necessary wav format. I'll also have to learn how to run Elias' code and Octave.
Perhaps the better way to do this would be for me to just email my impulse files to Elias for processing? I still want to learn how to use Elias' code but I'm interested is seen the results. Still I have to set up the woofer systems which will take a little time. I'm a little busy at the moment.
HolmImpulse is very handy for creating *.wav files from tabular impulse files using the Import/Export functions.
One thing to watch out for is the Holm assumes "," or "." are decimal point.
The Holm recognized column separators are semicolon, space, or tab.
Attachments
Last edited:
I read the first of your referenced papers on MTF at LFs, I was not impressed. Much is being simply assumed in that paper that I am not sure he has the right to assume. Namely the relationship between the MTF and perception at LFs - this has not be established to my knowledge.
Further, the arguments for/against modal density and the perception of LF sound in ever larger rooms is seriously flawed. (A reverb chamber is not a useful reference regarding modal density since its poor performance is a HF thing not a LF thing.
On thinking about Mr. Farenza's other paper, where he concludes that modal decay rate is the main correlate to perception, I am disturbed by several factors. If there is a solid Fourier Transform relationship between the frequency response and the time/impulse response in these systems, then the decay rate cannot be independent of the frequecy response. So if two systems have comparable flatness in their frequency response then they have to have comparable impulse responses and decay time - unless the system is seriously non-MP.
I am concerned that the "answer" here is being assumed to be correct and the data is used to support this answer rather than doing a proper design of experiments to determine if the hypothesis is truely being supported.
In some studies that I did I found that it would be desirable to have a long impulse response at LFs as long as this reverberation tail was not coherent. In thinking about this I do not see how the data shown in Mr. Farenza's latest paper can rule this out. It may well be that modal interaction is the more important factor and not modal decay as he concludes. In any "passive" system there can only be the number of modes that are actually in the room, but in an electronic one the electronics can add in "vrtual" modes if this is advantageous. Perhaps the preference for the electronically adjusted sub setups has to do with the addition of "virtual" modes - an increse in modal density and modal interaction - and not a reduction of modal decay.
Bottom line here is that I am not convinced by the work being presented that MTF has any strong bearing on LF perception as it clearly would at HFs. Nor do I see that a minimal LF impulse response is necessarily "prefered". Almost universally it is agreed that smoothing out the frequency response improves perception, so clearly that is what we want for the frequency magnitude of the rooms transfer function, but what is it that we want for the phase? Do we want a MP LF response? hence any and all systems with equally flat and smooth frequency response will have the same perception. Or, should the system be non-MP in which case the details of this phase becomes the critical factor. It seems to me that the data says that the later has to be the case, and I would tend to agree with that since this is exactly the same conclusion that I came to years ago. But then high MTF and a short impulse response would not be a good measure of LF performance.
A lot of guessing here and I am not sure about the answer, but I know that I am not convinced by what I have read as it seems to have too many holes.
Sure Dr. Geddes, it is a guess at this time, untill Elias or JohnK finish their independent experiments with sealed v/s cardioid. However to many and myself, dipole bass sounds more natural for music than a sealed. Do not know the scientific explanation...hopefully somebody will have have more objective data soon.
Perhaps, as the U-frame woofer has the closest structural resemblance to a one side open kickdrum, so do the acoustic properties come closest and so is the sound perceived from a U-frame woofer the closest to a kick drum ! So, "U frame woofer = electric kick drum" !!
Perhaps, as the U-frame woofer has the closest structural resemblance to a one side open kickdrum, so do the acoustic properties come closest and so is the sound perceived from a U-frame woofer the closest to a kick drum ! So, "U frame woofer = electric kick drum" !!
Last edited:
Yes indeed those fast bass lines in music are the ones that excell with dipoles.
And I think too that those whos primary goal is to have 115 dB at 20 Hz will find dipoles quite challenging.
Dipoles are a DIY thing. Commercially they tend to become more expensive than similar size reflex box.
We should be thankfull to be able to DIY these, and not depend on what's available on the market
- Elias
Sounds very good !
While the program is not difficult to use at all, feel free to send me any IR file needed.
If you send me one file first, I could do a 'test run'
- Elias
Never mind. The purpose is not to impress but to find answers.
Remember that I'm not using MTF, as I indicated earlier and expressed my concern on its usefullness to the task. TMTF could be more usable, I dont know, But I decided to go on my own way with wavelets.
If everyone would expect to find ready answers from literature like you seem to do, there would be no progress. There is no literature looking at the problem at hand from this point of view as I'm doing.
It didn't occur to you that this is novel original research going on right here ?
- Elias
Fazenda found that "Fast paced bass guitar notes in a funk genre. Bass notes closely spaced in time" were "not particularly helpful in revealing perceptual differences". "Slow, individual bass notes with short attack and decay; defined and isolated bass drum hits" were.
Sounds very good !
While the program is not difficult to use at all, feel free to send me any IR file needed.
If you send me one file first, I could do a 'test run'
- Elias
Hello Elias, would your program run in scilab (http://www.scilab.org/) also? Or does is use functions only supported by Octave
It didn't occur to you that this is novel original research going on right here ?
- Elias
No, not really. You act as if you are the first person to consider this topic and sudenly you have found the answer that has alluded all others. No, this does not seem very likely to me.
It is not "research" until it can past the test of peer review. None of your has and only some of the rest and I am saying that I am not sure about some of that either.
This thread sounds like "I know that dipoles sound best, therefor that technique which supports this conclusion provides the correct metric." That's not science. I know of no data that supports the assumption.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.