Here's another opinion. I started out measuring as a complete novice. Initially, Holm was a lot easier to understand compared to Arta. With Arta, I was not sure what I was doing, not sure if the settings were correct and so on. But as I started learning, and getting more experience, I found Arta to be the more accurate and powerful tool. Holm is still great, but Arta gives you complete control. It also has many other measurements that Holm simply cannot make: CSD, ETC, RT, burst decay, directivity sonogram (not free), just to name a few. It can also do spectrum analysis (measure your amps). And not to mention Limp, which can measure impedance and other free air parameters.
If you step back and look at Arta, it is one of those rare software that has so much functionality, built in a user-friendly interface (well, once you get the hang of it
), and in way that everything works, and works well. And on top of that, nearly all the functions are free, except you can't save the file. You really can't ask for more. It makes many paid software, even those that are not audio-related, look archaic and odd. Truly great piece of software.
If you step back and look at Arta, it is one of those rare software that has so much functionality, built in a user-friendly interface (well, once you get the hang of it
), and in way that everything works, and works well. And on top of that, nearly all the functions are free, except you can't save the file. You really can't ask for more. It makes many paid software, even those that are not audio-related, look archaic and odd. Truly great piece of software.
ETC is just another way of looking at the impulse response and an inferior one IMO. Holm does do a log display of the impulse response which does show the lower level things that can be hard to see on a linear plot. ETC is no magic way of showing anything - its just an integration which does some smoothing. I prefer to look at unsmoothed data thank you.
That's wrong as far as I know. Peter D'antonio has proven that ETC is always correct, when used for acoustics and reflections, while impulse response isn't necessarily. There's a reason why ETC is always used by acousticians.
Well, I think free is good enough for a hobbyist like me. It lets you export images of all measurements.
For a princely sum of $125, they offer full functionality. Peanuts for professionals.
Curious comment coming from you, given that your lively-hood appears to be based on designing and building speakers. Would you expect all your other tools used in developing speakers to be free as well ?
That's not to say that there aren't some excellent free software tools for speaker design/measurement, especially for DIY'ers, in fact there are many out there, but I think ARTA is one of those rare tools that has such a tremendous bang for the buck.
I was so pleased to discover it and find it fitted my needs so well (after trying and discarding many programs over the years) that I managed to put the money aside to register it despite being unemployed and financially hard up at the time...(and only needing it for Hobby/DIY purposes)
If I was doing this stuff for a living I wouldn't even stop to think about buying it for the functionality it offers, if only to support the author.
Arta is a very good piece of software, but some will be post-processing groups of measurements making this an early road-block.
I have used Limp and it is convenient, I'd use it again.
Edit: and I use Arta for centring the mic for listening position measurements, measuring excess group delay, and any single session work that requires large numbers of overlays.
Last edited:
I never said that ETC was wrong only that it wasn't always a good way to look at things. And the impulse response, if done correctly is also always correct. They are simply two different ways of looking at the same thing, but the unsmoothed impulse response shows more resolution than the smoothed and phase blind energy equivalent.
And its completly inaccurate to say that ETC is "always" used by acousticians - to wit my comments.
Appearances can be deceiving, but yes, I could buy anything that I wanted or needed, but I don't waste money either. I don't need or want ARTA, Holm does precisely what I want exactly how I want it. The rest of what I do is all custom software and there are no packages in the market that do what I do.
A few comments on impulse response versus ETC.
1) In what way is an ETC "smoothed" ? If it shows both the amplitude and exact time delay of a reflection, isn't that what we want ? Just how much resolution do we need in the time domain for a room reflection anyway ? Would it matter if the reflection arrives at 4.1ms instead of 4ms ?
2) Of what value is phase information in room reflections, beyond perhaps the first early reflection ? At higher frequencies second and subsequent reflections will have a more or less chaotic phase relationship to the original signal, and even your room criteria (no significant reflections in the first 10ms) doesn't say anything about phase.
The important information is what time delay does each reflection occur at, and how many dB down is it. ETC answers this at a glance, impulse response does not.
3) Yes technically the impulse response contains all the phase data, but it is not presented in a form that can be visually extracted at a glance in any meaningful way - either the phase shift vs frequency of a single impulse or relative phase shift between the first impulse and a reflections impulse. Even if the phase data was important we couldn't get it by just looking at a raw impulse response, we would first have to transform it into another visual representation.
4) Impulse responses normally have a linear amplitude axis, somewhat limiting their usefulness for looking at reflections that could be 15-30 dB down. (None of the software I checked could display an impulse response on a logarithmic axis, and I'm not sure that it would make sense anyway) ETC's normally have a logarithmic amplitude axis which gives us the data we need directly with a single glance.
5) ETC's can be either wide-band, allowing us to see the overall level of reflection, or (with the right software, like ARTA) narrow-band, allowing us to examine the level of reflections from a surface in different specific frequency ranges. This allows us to directly compare the effectiveness of speaker directivity, absorption and diffusion at different frequencies rather than just looking at the overall level.
An impulse response can't do this because we can't visually distinguish the relative amplitudes of high and low frequencies of the reflections, the visual representation gives a high priority to high frequency reflections but makes reflections with the high frequencies rolled off visually almost invisible amongst the "noise" of multiple reflections.
In every way that matters an ETC is the ideal form to display room reflection data, the impulse response is just the raw data which has not yet been processed into a form that can be interpreted in a useful way.
Last edited:
A good tutorial article here by Keele.
http://www.xlrtechs.com/dbkeele.com/PDF/Keele (1992-10 AES Preprint) - Analytic Impulse and ETC.pdf
You can think of ETC as the log magnitude plot of the impulse, although it is a little more complex than that (the RMS sum of impulse and differentiated impulse, to reveal true energy).
It may or may not include time smoothing. You can plot it unfiltered, which will emphasize the top Octave, or filter it as you like.
Phase of reflections? Do we mean time delay from the initial impulse or are we talking about some phase response of the issolated reflection? (All we care about is delay re. the direct sound.)
In the end we are looking for a metric that reveals the strong reflections in the overall decay so that we can compare that to the numerous subjective studies regarding their audibility.
David S.
http://www.xlrtechs.com/dbkeele.com/PDF/Keele (1992-10 AES Preprint) - Analytic Impulse and ETC.pdf
You can think of ETC as the log magnitude plot of the impulse, although it is a little more complex than that (the RMS sum of impulse and differentiated impulse, to reveal true energy).
It may or may not include time smoothing. You can plot it unfiltered, which will emphasize the top Octave, or filter it as you like.
Phase of reflections? Do we mean time delay from the initial impulse or are we talking about some phase response of the issolated reflection? (All we care about is delay re. the direct sound.)
In the end we are looking for a metric that reveals the strong reflections in the overall decay so that we can compare that to the numerous subjective studies regarding their audibility.
David S.
Energy is the time integral of the instantaneous power. The power can be determined as the product of two conjugate variable in a physical problem such as the voltage and current, pressure and velocity, etc. A way to estimate this power is to take the impulse response and its dirivative, as these represent approximations to the conjugate variables of the problem, and multiple. This product is the instantaneous power, but to get the energy we must integrate that result. Finding the RMS sum is a short simplification of this, but there is still the issue of how wide this sume is.
The span of this integral is important becasue the energy estimate is only accurate if several cycles of a given frequency have occured. So a short integration yields valid HFs but not for low frequencies. To broaden the applicability of this "energy" estmate we need a fairly long time average. This then smooths the data. The smoother it is the more valid the LFs are and the less it is smoothed the less valid the LFs are. Its a tradeoff linked directly to the uncertainty principle.
The impulse response has no such tradeoffs. It is what it is.
Last edited:
Hi Earl, Dave all,
I believe the ETC display Arta and other impulse based system show is essentially the same as what the TDS / TEF machine shows when one does an ETC measurement.
In Heyser’s view, the ETC was composed of both the impulse response (the real part) and the doublet response (the imaginary part) and represented the instantaneous energy as well as indicating Energy Vs Time.
That display is generated by using a linear sine sweep and then multiplying the mic and source signals. An FFT is done on the output which is the Sum and difference of the input signals (sum is discarded) and then the horizontal scale changed to show Time instead of frequency.
With a fixed sweep rate linear sweep, the first returning signal is always X behind in frequency and so it’s Time can be fixed and the phase detector given a place to start with.
For my use, the ETC is not useful for making a crossover alignment because it is based on a linear sweep and so the time information leans strongly on the high side of the systems response.
One can do an ETC through a pink filter to make all octaves equally weighted but that is not that useful either.
For room work, that might be more useful but I don’t do rooms and when I do, I use what I have..
This is kind of old but it shows the Heyser spiral, the 3d view of the impulse, doublet responses (p 138) and the connection between the different ways of seeing the signal (p 136).
I don’t think the room spirals are that informative but one of a speaker alone is more interesting. Page 139 shows the ETC (the entire energy envelope) with the Log Impulse squared.
It seems like the ETC and Impulse response are part of the same family as well as the largely forgotten imaginary friend haha.
Fwiw, it was Dick’s hope that eventually the measurement process would include both the air pressure and velocity components. Hard to believe he passed away 25 years ago now.
Fig 6-7
http://www.focalpress.com/uploadedFiles/Books/Book_Media/Audio/9780240808307.pdf
Best,
Tom
I believe the ETC display Arta and other impulse based system show is essentially the same as what the TDS / TEF machine shows when one does an ETC measurement.
In Heyser’s view, the ETC was composed of both the impulse response (the real part) and the doublet response (the imaginary part) and represented the instantaneous energy as well as indicating Energy Vs Time.
That display is generated by using a linear sine sweep and then multiplying the mic and source signals. An FFT is done on the output which is the Sum and difference of the input signals (sum is discarded) and then the horizontal scale changed to show Time instead of frequency.
With a fixed sweep rate linear sweep, the first returning signal is always X behind in frequency and so it’s Time can be fixed and the phase detector given a place to start with.
For my use, the ETC is not useful for making a crossover alignment because it is based on a linear sweep and so the time information leans strongly on the high side of the systems response.
One can do an ETC through a pink filter to make all octaves equally weighted but that is not that useful either.
For room work, that might be more useful but I don’t do rooms and when I do, I use what I have..
This is kind of old but it shows the Heyser spiral, the 3d view of the impulse, doublet responses (p 138) and the connection between the different ways of seeing the signal (p 136).
I don’t think the room spirals are that informative but one of a speaker alone is more interesting. Page 139 shows the ETC (the entire energy envelope) with the Log Impulse squared.
It seems like the ETC and Impulse response are part of the same family as well as the largely forgotten imaginary friend haha.
Fwiw, it was Dick’s hope that eventually the measurement process would include both the air pressure and velocity components. Hard to believe he passed away 25 years ago now.
Fig 6-7
http://www.focalpress.com/uploadedFiles/Books/Book_Media/Audio/9780240808307.pdf
Best,
Tom
He dropped by KEF when I was there and gave our group an impromptu lecture on many topics. Fascinating as I recall. Must have been around 1985?
David
Just FWIW, I modeled the bare BBC control room in CARA software to look at the reflections. CARA gives me a nice animation - alas not exportable to a universal format that can be posted here.
Next I'll build the BBC room with the diffusion panels. Gonna take awhile, as I'm both busy and slow. I'm very interested to see the differences.
Next I'll build the BBC room with the diffusion panels. Gonna take awhile, as I'm both busy and slow.
I'm very interested to see the differences.
Hi Tom
You have to understand that Heysers work was discredited sometime after he died when the AES would not publish his collected works. This is because the work was sent out to several renown mathematicians for review (not AES normal reviewers) and they all concluded that the work was not mathematically correct, so the AES declined to publish it.
Since the velocity is a conjugate variable to the pressure - its the spatial derivative or gradient of the pressure - the derivative of the impulse, Heysers "imaginary" part is a rough approximation of the velocity. Thus the product of the real and imaginary parts are a rough estimate of the intensity and hence the power per unit area. This has to be integrated over time to get the energy.
What a lot of acousticians looked at was using two microphones and directly measuring the pressure (the sum of the two) and the velocity (the difference) and multiply to get the power impulse or intensity. But this measure is directional. To get a non-direction signal we need four microphones and now things are getting tough to handle. However, some architectural firms do use four mics to show the direction of individual reflections.
Heyser came to Penn State and gave a talk while I was there. It was not well recieved.
Hi Earl
I am reminded that the dead are at somewhat of a disadvantage so far as defending / explaining theory’s against a live outspoken foe especially if seeking the lime light.
I recall that time and would add that part of that appeared to be partially a political thing as well with a rift between AES and Synaudcon factions, with Heyser being more in the Synaudcon camp and the TDS process being more like Beta tape (narrowly licensed but a little better than VHS).
After having a chance to poke around in the Heyser archive / museum at Columbia College a couple times, reading about some of what he accomplished at the Jet Propulsion Labs using some of the same techniques and reading some of his un-published work, I am left with the impression he was a brilliant hands on experimenter who’s world was actually centered in RF / electronics with hifi being a hobby.
In contrast to the other work he did, I believe only a politically or ego driven campaign could have discredited him / his entire body of work with such a broad brush as it has at the AES.
Even now, the TDS process remains my most reliable tool for measuring a drivers acoustic phase and magnitude with great noise immunity and was the basis of several acoustic imaging systems used in medicine and geology.
My personal experience with AES also eventually suggested it was not what it appeared and after a decade or so I dropped my membership.
I had been under the impression it was about “new stuff” discovery and invention. After presenting on the Servodriven subwoofer, the elimination of power compression and demonstrating a full range rotary loudspeaker and a couple others, there was never any interest in publishing any of them in spite of being “new transducers or systems” and usually requiring an extended question and answer time.
It was suggested back then if we were a supporter that things might be different and for some reason that really ticked me off at a core level.
It was at one of those old conventions I saw your paper on Bandpass speakers and then met you. I remember being scared when you stood up to ask a question in the Servomotor subwoofer paper but thankfully it was an easy one. Not having an academic background, just presenting was really stressful as those were my first presentations, thank heaven they had a pitcher of water handy I needed it.
I met Dick once at an AES convention back then too, Don and Carolyn were visiting our booth and suggested I walk with them. The chance to escape booth duty turned into a dinner with them, Dick, Gene Patronis and some others. I said about three words the entire night and then only in response to a softball question.
Dick was like a very few people I have met, in fact your one as well, people that think / see math concepts, that walk through a hedge of thorny math like it was manicured bent grass.
I am a visual person, for me working through and then understanding the math is usually near the end, after I can measure it, visualize how something works. As a result, I am near the bottom of the pile in an ability to judge his math or how much is academically rigorous.
I have a number of his papers that if it could, I would love to ask “what did you mean by this” but obviously I can’t ask, only wonder what he meant and wonder how the heck he got there so long ago.
Best,
Tom
I am reminded that the dead are at somewhat of a disadvantage so far as defending / explaining theory’s against a live outspoken foe especially if seeking the lime light.
I recall that time and would add that part of that appeared to be partially a political thing as well with a rift between AES and Synaudcon factions, with Heyser being more in the Synaudcon camp and the TDS process being more like Beta tape (narrowly licensed but a little better than VHS).
After having a chance to poke around in the Heyser archive / museum at Columbia College a couple times, reading about some of what he accomplished at the Jet Propulsion Labs using some of the same techniques and reading some of his un-published work, I am left with the impression he was a brilliant hands on experimenter who’s world was actually centered in RF / electronics with hifi being a hobby.
In contrast to the other work he did, I believe only a politically or ego driven campaign could have discredited him / his entire body of work with such a broad brush as it has at the AES.
Even now, the TDS process remains my most reliable tool for measuring a drivers acoustic phase and magnitude with great noise immunity and was the basis of several acoustic imaging systems used in medicine and geology.
My personal experience with AES also eventually suggested it was not what it appeared and after a decade or so I dropped my membership.
I had been under the impression it was about “new stuff” discovery and invention. After presenting on the Servodriven subwoofer, the elimination of power compression and demonstrating a full range rotary loudspeaker and a couple others, there was never any interest in publishing any of them in spite of being “new transducers or systems” and usually requiring an extended question and answer time.
It was suggested back then if we were a supporter that things might be different and for some reason that really ticked me off at a core level.
It was at one of those old conventions I saw your paper on Bandpass speakers and then met you. I remember being scared when you stood up to ask a question in the Servomotor subwoofer paper but thankfully it was an easy one. Not having an academic background, just presenting was really stressful as those were my first presentations, thank heaven they had a pitcher of water handy I needed it.
I met Dick once at an AES convention back then too, Don and Carolyn were visiting our booth and suggested I walk with them. The chance to escape booth duty turned into a dinner with them, Dick, Gene Patronis and some others. I said about three words the entire night and then only in response to a softball question.
Dick was like a very few people I have met, in fact your one as well, people that think / see math concepts, that walk through a hedge of thorny math like it was manicured bent grass.
I am a visual person, for me working through and then understanding the math is usually near the end, after I can measure it, visualize how something works. As a result, I am near the bottom of the pile in an ability to judge his math or how much is academically rigorous.
I have a number of his papers that if it could, I would love to ask “what did you mean by this” but obviously I can’t ask, only wonder what he meant and wonder how the heck he got there so long ago.
Best,
Tom
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.