Hello

I started this thread to have more information about how to doing this measurement and how to interpreter the graph. I hope maybe can became a sticky

I used the search function but unfortunately ETC is a too common acronym (etcetera), so I haven't found many information (even using impulse or envelope time curve)

What I do is a sweep from 20 to 20.000 with room eq and I can visualize the "impulse tab" to see the ETC

It's that it? I read the ETC is an impulse, but I haven't figured out how to play an impulse in room eq, seems room eq does a sweep and get every information, measurement, from that sweep

So, I'd like to know if there are more correct way to do an ETC or if that way is reliable

Second question is: how to interpreter the ETC graph? What should I see?

Links to other tutorial are welcome, I imagine this topic has been already covered, I just can't find it using search or google, maybe because of my searching skill, but I think because of the popularity of the acronym ETC

Thank you all

A good start:

Sound system engineering - Google Böcker

elan, you asked similar questions in passing in a PM. I will make a brief and simple answer here. There will be many who have deeper, mathematical and so on, understandings. If so please help out by adding to the understanding or correcting it, if I have oversimplified to the point of misconception.

It is clear to me that you know what an Impulse is, but let me tie it down.
A perfect Impulse would be all frequencies, 20-20K, starting instantly and simultaneously. Comically a balloon bursting is pretty good and can actually be used. Or a starter pistol shot.

A Sine Sweep is curiously also 20-20K. I am sure you will see my drift already. So the space is subjected to all the frequencies of interest, just spread over time. And the rest is just Arithmetic....

When you take a measurement in REW FM etc. you are actually recording the response of the room to the stimulus. The Recording is called an Impulse Response.
The software can extract this room response equally from an actual Impulse or from a Sine Sweep or even some other forms of stimulus.

I am not sure about REW, but try recording a handclap, use the Import or equivalent function and see what you get.....

DD

seems hard to believe a sine sweep would be useful for a time based measurement.

ok Dan, simple and effective :P
I imagined it was as you said, I just wanted a confirmation

So, now remains only the second question:

How to interpret an ETC?
I mean, if I understand correctly I have to look at the "spikes" and some of them are the resonances.. then see at what ms they happen and calculate the distance using this Conversion of time difference and sound path distance or length millimeters time of arrival calculate calculation noise sound wave in air calculator - sengpielaudio Sengpiel Berlin

Then I take a string of the length I got from the calculator and then I can find where the resonance happen, what causes the resonance

What isn't clear is how to identify which are the important resonances or the type of resonances, in the ETC

Thanks for the patience, I haven't found this explained clearly

ok, so how you do an ETC with room eq?

It’s time to put the use of an ‘impulsive’ signal to bed as an adequate source for measurement and analysis. There is good reason their use is not commonplace despite the romantic attitude some have of them.

As an impulse approaches the ideal of a Dirac impulse (Dirac delta function) of zero time and infinite amplitude, the bandwidth of the signal in the frequency domain becomes wider , following Bw = 1/T.

Thus, an impulse with a T= 50 usec becomes a bandwidth of 20 kHz or (1/.00005) = 20,000 Hz.

Seemingly, this signal would be an ideal source if it were not for a fatal flaw in energy accounting. As the time interval shortens, the amplitude must increase in order to maintain the same energy level. By the time a useful time interval is achieved in terms of useful bandwidth being generated, the signal’s amplitude is so high as to overdrive almost any test device. If the amplitude is reduced, so is the total energy content and the resultant signal-to-noise situation becomes so severe that meaningful measurements are not possible in any kind of normal ‘in situ’ environment. In artificially quiet environments, impulse measurements can be instructive tutorial examples of sources for Fourier transformation of signals. Compared to swept sine wave techniques, impulse signals suffer roughly an 86 dB disadvantage.

The impulse response is pure amplitude and subject to non-linearity. The swept sine wave is full power at all frequencies and has the immense benefit of being sinusoidal, hence traceable through complex systems for time, phase, and amplitude.


Additionally, using the impulse response alone to analyze behavior is likewise fundamentally flawed in that it takes into consideration only the real domain and an FFT of the impulse only generates the coincident response while utterly failing to account for the imaginary doublet response from which the FFT would generate the imaginary quadrature response in the frequency domain. There is nothing imaginary about the so-called imaginary part! And both are absolutely necessary if we are to see all of the information contained in the time response of the signal.

This point has been aptly demonstrated by Peter D’Antonio who has presented measurements featuring significant energy returns that only appeared in the doublet response and not at all in the impulse response.

Unfortunately, the oft cited claim that the impulse response contains all of the necessary information is a bit overly simplified. What is often ignores is the need to apply the Hilbert transform to the impulse response in order to obtain the doublet which along with the impulse is squared and then summed, and then the square root taken and displayed with logarithmic amplitude as an envelope time curve.

Yachting canons, hand claps, potato guns, starter pistols and the like do not exhibit a uniform maximal intensity power response across the entire bandwidth. They should be left to their original uses.

Chris, Thank you for that clarification!


Thus, if I understand you correctly, assuming hardware loopback (automatic device correction) is used to account for hardware latency/propagation delay, if one does not apply FM’s “minimum phase” origin translation, then each energy peak displayed in the ETC will correspond to the actual Time Of Flight (TOF) from the acoustical origin (source speaker) to the measurement mic.

Folks, for our acoustic purposes here, that is exactly what we need to be using.

Normalization relative to the direct arrival point, Td, where time=0 is assigned to the Td, is NOT useful for our particular application here. While for delay settings and for a few alternative purposes it is very useful feature, is it NOT useful for our purposes in this application.

So, if you are using FuzzMeasure, (and if I understand the implementation properly),please be sure to enable the “Automatic device correction” (hardware loopback in order to compensate for hardware latency/propagation delay) and do Not use the “ minimum phase” setting. If this configuration is followed, the time of each energy arrival will correspond to the actual time of flight.

In order to determine the actual distance of travel for each direct and indirect specular energy path, one multiplies the actual TOF x 1.13 ft/ms or .344 m/ms for the distance in feet or meters respectively. This will be useful if one is not using one of the blocking methods for identifying the correlation between ETC and thee various specular paths.

If one is not empirically determining the path(s) of the indirect signal(s) by blocking, and want to know the actual distance of the indirect signal, you need to know the Total time of Flight (TOF).

This information may make the process a little easier to visualize as you can now know the actual time and the actual distance traveled from acoustic source (your speaker) to the measuring mic.

Thus if you choose to determine the specular energy paths in the most basic and accurate manner, you can, depending upon the amount of separation in time of the various reflections, more easily determine which boundary surface correlates to each energy spike. And from there, (omitting a few mechanical steps*) you can determine precisely the location of the center of the incident region.

And, assuming one knows the acoustical model** they are working to satisfy; one indeed knows where to place treatment. The type of treatment is then determined by the effect one desires to create.

Absorption will damp the reflection.
Reflection will redirect the energy which will effectively cause incidence later in time at other location(s) - in other words, it will both reduce the gain of the spike and it will effectively be 'moved' to a later time.
Diffusion will do two things. One, it will decrease the gain of the spike. And two, it will break the primary reflection into 'smaller' reflections of lower gain and spread them out in time. Thus you will have a nesting of distributed lower gain spikes, generally with the distribution in time being skewed to a later time.
(See graphics below)

Thus, if you know the target acoustical response desired, reading and interpreting the ETC correctly will indeed provide information as to the precise point of incidence, and this point is where treatment is applied to mitigate said energy in the manner desired. And knowing the acoustical response desired, you can appropriately choose which kind of treatment is useful at the location.

Note, I say what type of treatment may be useful rather than what is necessarily best. The reason is that there are often multiple ways to achieve similar results depending upon the context of your space. For instance, if your acoustical model is that of a NE room, then you will most generally want to use absorption to reduce the reflection. On the other hand, if you are building according to the LEDE model, and you want to preserve the energy while simultaneously controlling its dispersion; then you may want to employ either reflection or diffusion in order to create the acoustical response appropriate for that portion of the ETC response. But a comparison of the actual ETC with the acoustical model template will help you to determine what behavior is optimal with respect to time and gain for the given incident point in the room.

Does that make sense? The ETC provides a total picture of the specular response in the room - from early arrivals to the 'last' of the energy, be it totally damped or a decaying diffuse soundfield. And this is all done with respect to time.

The ETC allows you to see exactly what kind of energy distribution you have currently, and allows you to select and precisely place the treatment you have chosen in order to create the effect you desire - be it damping, redirection, or diffusion. It also allows you, upon repeating the measurement, to see the precise impact the positioning of the chosen type of treatment has. from this you may be satisfied, or you may want to further refine the positioning in order to insure the proper response is accomplished.

Oh, and one more important point here. It will also show you if, and to the degree, that your treatment does not act completely in the manner you suspect. In particular, this is most common with absorption, which to many folks surprise, will often exhibit a stronger degree of reflection than anticipated - especially if the angle of incidence at the boundary is great. you can also determine the actual degree of diffusion versus scattering a diffusor or a scatterer such as a poly-cylinder exhibits. With this information, you may decide to modify or use another type of treatment if , for instance, the resulting reflections are not sufficiently diffuse.

But in any case, if one becomes proficient in using the ETC, the days of blindly assuming a treatment based simply by virtue of its name, performs exactly and solely as one expects, should be over. you will discover that absorbers have a reflective quality. And that diffusors exhibit an absorptive component (often more than one would like if your goal is to diffuse and retain said energy!) and that they may also act as reflectors (especially is the incident signal is perpendicular to the unit). In other words, you will not only know what is happening within the room, but you will quickly learn a great deal about the real, as opposed to ideal, behavior of the various treatments.

And with this combined knowledge of both what is happening in the room, having the information of where to place treatment appropriate to your acoustical response design goals, knowing what specific affects your choice of treatment actually achieves - while becoming aware of any residual artifacts of the treatments, you will be well on your way to creating the response you desire....


-----------------------------------------------------------------------------------------------------------------------

* …assuming a speaker system incorporating DSP is not being used, the hardware loopback (automatic hardware correction) is adequate. If, however, a speaker featuring DSP , such as the Renkus-Heinz Iconyx for example, the latency of the DSP will have to be accounted for separately and in addition to what has been mentioned. This should effect few people..

** there are a number of ways to identify the paths, ranging from very basic to more sophisticated. With each increment, the amount of physical work is diminished as well as the time required. With a bit of practice, pattern recognition, ans in some cases an investment in a more sophisticated measurement platform, the rate and precision can increase (until you are dealing with a 3D polar ETC where you place your cursor on each point displayed on the display graph and the 3space X,Y,&Z coordinates are displayed allowing you to simply replace the measurement mic with a laser pointer and aligning the laser at the generated coordinates and having the laser point precisely to the optimal incident point on the boundary.)

Until then, here are two easy techniques that are both very precise and utilize the time and distance information provided by the ETC itself.

1.) - the most basic mechanical method is the string method.

Quite frankly, I don't expect anyone to do this more than a handful of times until you can more easily visualize the process, but it illustrates the concept very simply and effectively - if a bit awkwardly!

From the total time of travel, you calculate the distance of travel (TOF X 1.13'/ms or .344m/ms). You might want to leave a few inches on each end to hang on to, but mark the precise endpoints corresponding to the distance. Find a few friends who you will not mind being a bit worn out by the process and have then each hold one end of the string, one end with the point marked on the string precisely located where the measurement mic capsule would be (don't move the mic!!!) and the other end placed in the center of the source speaker. Forget the tweeter stuff, as the ETC measures the total energy, and there is much more energy content in the low-mids and mods than in the tweeter. Besides, if you want to determine the actual acoustic origin of the speaker, the ETC can be used to identify this as well! But that’s another exercise for some other time!. Now, with the endpoints firmly located, at one point in the body of the string, extend the loop body out and see what boundaries/surfaces you can tangentially touch with the string being stretched ‘taut’ The point you can touch, is the point of incidence indicated by that particular spike in the ETC. Note the incident spot on the boundary. Repeat for the other energy returns.

Easy, but I suspect you will quickly tire of this – as will your friends who will want to be doing something a bit more exciting!


2.) - Alternative Methods:.

“Although it’s often easy to hear where a reflection is coming from by playing a pulsed signal and cupping your hand to your ears, sometimes it’s necessary to employ measurement systems to do it. Looking at an ETC or a log-squared impulse response, we can see a reflection as a spike that sticks out above the reverberation decay.

Place the cursor on the spike and note the arrival time, or set that time as a reference for difference measurements. Now move the mic a small distance, say 6 inches, in the direction you think the reflection is coming from. Take another measurement. If the reflection is earlier in time, you’ve moved towards it. If it is later in time, you’ve moved away from it. If it didn’t change much, you’ve moved sideways across it.

Try another direction and see which way the reflection moves. Also try moving the mic up or down. A lateral reflection from a side wall won’t change much, but a ceiling reflection will change a lot. Also be sure you are looking at absolute time units, not time relative to the direct arrival. The direct arrival will be moving around as well.

[I]Another way to determine arrival direction is to block the microphone’s view in a certain direction with a sound absorbent barrier, such as a sheet of Sonex. When you block the offending reflection its spike will go away (See blocking technique illustration below.)

*** The acoustical model is the acoustical response of the room. Examples are the Non-Environment room, and the LEDE room. These are defined most completely in terms of their ETC response. If one can interpret the ETC in terms of arrival time, gain, and the spatial temporal energy density indicated by the spacing and gain of the energy arrivals, it is easy to use the ETC as a template for the creation/recreation of the room model.


ETC & Treatment: Absorption, Reflection and Diffusion:







Blocking Techniques for isolating/identifying reflections:






Acoustical Room Models expressed as stylized ETCs:






Note: the ETC can be used for a myriad number of applications. The above mentioned is just one of many.





And a few more thoughts based on a few comments above.

There seems to continue to be some persistence in looking at specular responses in the frequency domain and worrying about EQ.

If you address the time domain issues in the specular region, you don't have to worry about the frequency domain or EQ! EQ is appropriate ONLY for adjusting the direct signal that comes out of the speaker - NOT the speaker-room interaction. And if you do choose to go the route of EQ in the room, you might want to ONLY consider it for modal control. But I am not a fan of that either.

The ETC and the attendant specular reflections address all of the behavior above room modes. The ETC and specular behavior is addressed in the time domain - not the frequency domain.

Resonance is a LF modal issue. You examine modes with a ~0-300 Hz FR/waterfall plot. You only 'need' the frequency response for the waterfall. After (above) that, worry about the time domain. If you resolve those issues in the time domain, the issues that may manifest themselves in the frequency domain are minimized.

Much appreciated, SAC. Very informative.

I have a question regarding the transition from modal to specular behavior. I've seen the terms "Davis" and "Schroeder" frequency used to describe this this transition point. I've also read that trying to calculate this frequency is not worth one's time. I've further read that in order to distinguish a modal peak from a specular superimposition, one relies upon the waterfall plot: modal frequency peaks "ring", as Ethan puts it, showing longer decay times in the form of waterfall "peaks" (time axis), vs specular frequency peaks, which do not ring.

My question is this: in order to determine where this transition occurs, do you look at waterfall plots for increasingly higher frequency peaks (in the 300-400hz range for small rooms) until no more corresponding "ringing" occurs? I'm guessing that the "transition zone" in a room includes frequency peaks whose decay times fall within a certain range. Can you describe that point in terms of graph analysis at which one should stop looking at waterfall data and switch to ETC?

My application to the answer of this question: if designing a diffuser, how do I analyze the graphs to determine its low frequency cutoff if trying to deal with all "diffusable" (specular) energy?

Thanks again, this really helps.
John

Thank you SAC
And thanks to have used the metric system :P (for the formula to find the distance of the resonance)

OK…Modes.


You do not need to calculate the Schroeder/Davis ‘critical’ frequency.

Don't 'over think' this too much. It is sufficient to be aware of the nature of the behaviors. Seldom are you going to run into a room topology that will cause a significant problem in the region. And the waterfall and ETC are sufficient to 'address' even this.

Modal activity is related to the room dimension being equal to or larger than the wavelength, thus the larger the dimension, the lower the mode that can be supported.
Thus, the smallest boundary dimension is typically from floor to ceiling. And this typically corresponds to the highest fundamental mode. If that is 8 foot, then the axial mode is normally in the region of ~72 Hz. The modal activity above that point consists primarily of harmonics and the coincidence of multiple mode harmonics clustered together. There are going to be modes in a small acoustical space. And as the modes increase in frequency as a basis of harmonics and coincidence, each of the contributing modal harmonics is lower in energy than their fundamental modes.

Thus, the critical frequency is going to be the point at which the dimensions of the space cease to be larger than the wavelengths. This is actually rather low and defined by the smallest dimension, as wavelengths shorter than that dimension will function as specular reflections. And since not all of the boundaries are the same dimension, there will be a mixed environment where both modes and specular reflections are present. And while the waterfall will display the fundamental modes and the several harmonics, the ETC displays energy with respect to time - it does not care what frequency - so all of the specular energy is accounted for.

Porous corner bass traps are not tuned. They are broadband. Now, depending upon the desired acoustical room response you may want to make them band limited in order for them to not absorb the mid and high frequencies.

As far as a convenient cutoff point. We suggest a waterfall display covering about 0-300 Hz - plenty sufficient to encompass the first few modal harmonics. Thus whatever we are looking for modally typically falls within that range. Broadband traps take over above that to control specular energy and are placed at positions of boundary incidence as identified by the ETC.

As far as diffusion is concerned, a good working tuned frequency is in the region of ~900-1000 Hz, placing the lower limit at ~450-500 Hz. For early reflections you are typically going to use broadband absorptive panels. I suggest a minimum of 4 inch thick panels – with up to 4” of spacing if you can afford the loss in real estate. Thus, their natural lower extension is sufficient to cover the transition zone, while the bass traps are normally sufficient to address the fundamentals modes. If they are not, you will probably want to explore tuned resonant absorbers.

You do not generally use diffusion for early first order reflections simply because in a small room, the energy return is too high in gain (and due to the typical close proximity, too specular and not sufficiently well mixed) that it will not satisfy the minimum 20 dB SPL down requirements of the ISD (relative to the direct arriving signal gain). Thus absorption is your friend in these spots.

If you recall the psycho-acoustical issues that are determinant regarding these ranges, unless one has the means to build incredibly complex high order diffusers (say from 3 to 7 foot deep each!) capable of providing both extremely well-mixed diffusion along with high degrees of gain control (and the fork lift necessary to move them within the small space – don’t forget to allow for ingress and egress of the fork lift! – a new design concern….) diffusion is not very appropriate for early first order specular control.
Thus diffusion is going to typically be used in the rear half of the space. And at this point we cease to address simple mode/reflection control and we are well on our way to implementing one of the acoustical response models which will play an increasing role in how the energy is directed and managed.

So, you needn’t calculate the Schroeder/Davis critical frequency. But you should be cognizant of how and why the behavior transitions from modal to specular based upon the relationship of boundary dimensions to wavelength.

And the frequency response/waterfall plots are useful for viewing modal behavior specifically because the wavelengths are large. And the benefit of the ETC is that you needn’t worry about frequency – as we are now looking at specular energy flow in the time domain. All specular energy, regardless of frequency. So, if you address the energy displayed in the ETC, there is no general need to EQ. And what EQ is done is done to the direct signal emanating from the speakers.

And if you successfully complete the tuning of the space, as the very last step, if you have the tools sufficient to identify possible minimum phase regions in the resulting room response, then you can apply PEQ. But ONLY to the minimum phase regions.

Excellent posts above!

http://www.acoustics.net/objects/pdf...e_farina02.pdf

Wonderful! Thanks!

I'd like to give a bit of background to propose a thought experiment, one that I'm almost sure has been carried out in real life. I hope that this at least tangentially relates to ETC and that Elan, the OP, will forgive my partial highjack of this thread.

I'm currently building a home theater and have been looking into various methods for treatment of first reflection points. The room is roughly 19 x 12 x 8 feet in L/W/H. I've read here that by the time one uses enough porous absorbers to deal with early reflections (in addition to those used for modal control), much if not most of the energy in a room has been eliminated.

I wanted to keep the room a bit more live and had the idea that multiple vertically oriented polycylindrical diffusors placed at first reflection points angled from the side walls could create vertical diffusion while horizontally angling early reflection energy away from the listener. This would allow for multiple additional reflections to occur before the sound ultimately reaches the listener, thereby decreasing the gain and increasing the arrival time of the initial first reflection. See drawing below.

Is this solution viable?

Thanks,
John

Excellent posts indeed, bookmarked, will read more than once I suspect.

A little housekeeping.
John, it is kinda hard to believe that a sine sweep can be used for time work.
This use of convolution becomes almost magical in the case of IR Reverbs, e.g. Altiverb.


Elan
I wouldn't call them resonances. The spikes in the ETC are reflections made visible. Consider a handclap. This will show up as a short very loud spike on an ETC. Then, hard surfaces nearby will reflect it. These individual reflections also be visible on the ETC. They will be later of course and this time lag can be used to locate them. The nearby surfaces may not reflect fully, plus distance attenuates, so the reflections can be quite weakened. If they are 15-20dB lower than the initial spike, we may ignore them.

SAC, great posts thanks. I just skipped through so I may have misread this, but if not this is meant simply FYI.
The Minimum Phase in FM is actually a menu command rather than a setting or preference. 'Create Minimum Phase Copy'
This seems to be of some benefit for some graphs, exporting 'cleaned up' versions etc. Good to know that it shouldn't be used when the Loopback is engaged.


DD

Ok, great, I'm starting understanding better about ETC
BTW here is a thread with a lot of measurements I done, would be great to have an opinion on real world measurement, other than theory, I mean I like to hear opinion that can guide my process

My Studio Room Tuning

Now, as you can see in some measurements, I have all reflections under -20 db

My goal is now finding the place for all needed absorbers, so if they are definitive I can build a frame, with slats over it, to make the room more live, but, considering the wood here costs A LOT (you don't even have an idea) I can't afford to build things and then change them, they should be quite definitive, so I'd like to know in general if I'm on a good way or if I have to add other absorbers to tame some dips and get a rid of the comb filtering, or simply change the position of something. For example my desk changes the response quite a lot, I'm trying to find the best placement for the desk and speakers, but sometimes I'm not so sure if a response is better to another or not, that's why I like opinion on it :P

I don't know in general if the ETC looks good, but it's under -20

Someone also knows how to "Create a minimum phase copy" in REW?

BTW great posts, a lot of informations in here :D

No problem, it's in topic
Interesting also, I just think maybe those panels should be angled in the opposite way, to the backwall

Whoa boys! before we go off the deep end...

We have discussed the ETC from the point of view of identifying the individual reflections and determining their gain and spatial/temporal qualitative characteristics.

We have mentioned early reflections very generally and referred to a minimum 20 dB SPL gain value relative to the direct arriving signal. (This is in reference to what is called he Initial Signal Delay(ISD) or Reflection Free Zone (RFZ) gap. this is a fundamental feature of one of the acoustical models.

But that is not all there is to room tuning!!!! The ETC is a tool, not the goal! You still do not have an acoustical room model to follow - that is what defines the behavior that you will use the tools and treatments to achieve!

Once you are familiar with the various room models, and evaluate your usage, needs, preferences and the limitations of the space, you then use the ETC as the primary tool guiding the means of adjusting the energy distribution within the space according the the desired acoustical room model!

If you look at those room models (ignoring the reverberant model which is not for listening rooms), only two of the models come close to typically used listening rooms. One is the Non-environment room (which is close but not necessarily identical to the anechoic model) and the other is the Reflection free Zone/LEDE model.

In all but the anechoic model, you have a termination to the Initial Signal Delay gap where the early arriving reflections are damped >= 20 dB SPL and you have a later laterally arriving exponentially decaying diffuse energy return.

The point being, the goal is not simply to render the room dead! We don't need a fancy tool to do that! The purpose of the tool as we are proposing it here is to surgically evaluate and treat specific anomalous issues.

Thus, once one understands and become proficient in the analysis of a space using the ETC, to then learn more about the various acoustical room models that you will use as your design template!

May I suggest, once the ETC is understood, taking a bit of time to familiarize oneself with the various acoustical models and learn the concept and rational behind them. Additionally you will want to familiarize yourself with the specific and actual behaviors of the various treatments. Then you can evaluate your application and space and determine which model is most appropriate for your use. And THEN you can use the ETC and the various treatments to tailor the response of your room to meet those requirements suggested by the chosen room model.

ok, I give up

tonight I burn my studio, everyone is invited

Thanks for the impulse vs. sweep link Lupo.

I'm doing my best to digest it. Lots of math. A few thoughts after my first read through:

It (and SAC) states that a sine sweep is preferred, as it "in a nutshell" disregards time based errors within the playback and capture system... among other things. Correct? If this is the case, typically (for diy'ers) playback and capture systems used for measuring are the same to be used after measuring in the listening/recording environment. Is it not beneficial to know whether or not your setup has these errors, so we can improve upon them? If playback is flawed at the speakers, how can we expect results from the room?

A sine sweep lasts what... say 10 seconds? How can the analysis detect reflections of a lower intensity through the direct signal that is still playing? I would think that say a 500hz reflection would be hard to detect under what may be a 600hz direct sound happening concurrently. Is the software so advanced that it compiles start times of each frequency and is able to pull reflections of that frequency from the ensuing "mess of sound" and after the fact realign all of those reflections onto a single start time? And is it this "compilation of frequency bands" and their averaging that makes the resulting analysis more precise? Understand I realize that ETC is not a frequency based measurement... thus why I have a hard time grasping why the sweep is desirable. The signal to noise aspect makes sense... I guess.

I'm not calling to question why this is done. Clearly more learned people have come to this conclusion for us. Just trying to "get it".

Measurement techniques is a large topic that is definitely far afield and beyond the scope of a thread directed towards the practical interpretation of the ETC.
Such questions in the applied realm are generally environment specific and will also vary dependent upon the configuration of the measurement platform abilities (i.e., weighted sweeps) and might be better addressed in the context of the measurement platform being used. (An analogy might be while exploring the process of adjusting a carburetor on a particular engine someone questioning the fundamental viability of he internal combustion engine! ;-))

Here we are assuming that one has gotten past the basic configuration of the the measurement platform and wish to know what a successful ETC response indicated.

There is not a one size fits all application either. That is why the sweep parameters are definable. There are also various types of swept signals ranging from sine waves to Galois sequences (MLS), to TDS, to various abilities to specify custom configurations....

If you examine the capabilities of the larger comprehensive packages you will quickly discover a very large range of available stimuli available along with many qualifications to the signal.

And the duration, weighting, etc of the sweep all have an effect as well.

Measurement techniques is a rich field in itself. I am not sure what benefit questioning the foundation of the tools plays here in this thread; although a knowledge of the function of the tools sufficient to operate them effectively is indeed a worthwhile virtue!!!

It could indeed be the subject of an extensive thread as there is much to be investigated! Just be prepared for the math that necessarily accompanies the subject.heh

As far as the basic viability of the approach within the narrow scope of our purposes here, when all of the major measurement and analysis packages incorporate this, ranging from REW, FM, RPlusD, up to such platforms as ARTA, Smaart, Sim, TEF, Room/WaveCapture, Easera/SysTune,B&K, etc., one can reasonably assume there must be 'something' to it...

Yes.
There is Convolution involved I believe.
Or in my view, Magick.

DD

thank you Dan.

Cute!

Müller & Massarani's paper is easier to swallow, you can find it at CiteSeerX — Transfer-Function Measurement with Sweeps DIRECTOR (use the "Cached" link, the original link is dead).

My understanding of the "ambechoic" room model, executed in Blackbird Studio C, creates an environment in which the arriving reflections are no less than 30db down from the initial signal, and decrease in amplitude at arrival times thereafter. See pages 13 and 24 in the link below:

http://www.cari-llc.com/pdfs/Mono2Surround.pdf

Studio C is quite large, certainly more so than a typical residential room. Can an "ambechoic" result, or something close to it, be achieved in such a room? If so, how might one go about designing/treating this room?

Thanks,
John

Blackbird seems to be the only example of Ambechoic... at least that I've seen. Those 3d qrds are like over 7' deep.

The mention of fork trucks needed for moving always comes up.

It also tends to be a surround model, though again, I'm not sure what constitutes "tends"... as this room is the only example I've seen presented.

Not quite....

The arriving signals are not 'reflections' in the traditional sense that you use the term. They are diffuse in the statistical sense. To the point of achieving a statistically reverberant space where there ceases to be attributable direction associated with the 'reflections'. In other words, the probability of the energy arriving from EVERY direction simultaneously and equally is satisfied.

You might want to take a look again at the diffusors required to accomplish this on page 22 and 27.

Then ask us if this is achievable on a practical level considering that each of the N-157 ceiling PRD diffusors are 7 foot deep featuring 24,336 block heights...and that each of the wall PRDs are "3’ deep amplitude modulated prime 181 and 769 1D primitive root sequences, using modulus 953 and 138,646 block heights!" (Considering that neither the diffusors nor the forklift required to move them fit through a standard door, you might also want to consider widening the doors a bit... ;-) )

Let's see, that's 7 vertical feet taken up with diffusion and 6 feet of width taken up by diffusors alone.

Considering the difficulty folks are experiencing at the thought of constructing an N=7 1D QRD, folks might want to start by getting the basics down first. This is rather like someone asking if its possible to obtain a 12 week classroom private pilots license and then asking how one then flies an SR-71 Blackbird for the hands on test component.

The only thing this topic has to do with the ETC is that the acoustic response model, like all others, is defined by an ETC response.

heh Fair enough. My question probably deserved its own thread, but I tend to take liberty with the word "tangent." Thanks anyway.

John

Very interesting guys, I did both of the tests by SAC (the string and the mic moving), i could identify (or confirm) the my desk was causing a reflection..
I have question about FMs ETC.
Ive noticed that when i drag my start timeslider (top left corner of the ETC), my frequency response graph changes, is there any relation with those, if so which one?

In my case i could assume that my first reflection was creating a null (if i drag the slider just over my first peak (a bit after actually, 4.74ms), the graph will show me a pronounced dip at 133hz.

But i might be wrong (probably am).

Is there then another way of knowing the frequency involved? If i do the string method for example, i will know what surface is involved, but the frequency? If the frequency is high, wouldn't be a multiple of it?

Finally as you can see on the graphs, i am assuming my first reflections are is at 4.74 but i am not sure, is it normal to have so many spikes? On my case, would i be considering: just over 5ms, 6, 6.5 and just over 8ms as my peaks only?

I am sorry if i offend anyone with my lack of knowledge, i've learned more today than in the last weeks.. so its a bit difficult to grasp all that entitles this new concept for me (ETC).
Thanks for your time

You have plenty of destructive reflections(at least ~5-7 discreet reflections) between 3.7ms and 4.2ms that must be addressed.
Also, please make two ETC measurements with one speaker (L/R) driven at a time...