Sound Design Live

Build Your Career As A Sound Engineer

Know Your Audio Analyzer Averages

By

After you take several measurements and average them together, what do you expect to see?

If these two measurements are averaged, what do you expect?

Is zero the average of -6dB and 6dB, or something else?

1 octave wide parametric filters at 1kHz

Here are four possible averages you may have guessed, depending on which audio analyzer you use.

tl;dr

  1. Know what kind of averaging your audio analyzer uses.
  2. Collecting more data is more important than the way it is averaged.

Here are demos from a selection of audio analyzers in alphabetical order.

Crosslite+ v2.0.0 8

Along with options for pre and post processing, Crosslite+ offers four different options.

“Arithmetic Average Complex” : Arithmetic mean in complex values.

“Quadratic Average Complex” : Quadratic average or RMS in complex values.

“Arithmetic Average Magnitude” : Arithmetic mean in real values in dB, with phase zeroed.

USER GUIDE CROSSOLITE REV 1.1

Does the magnitude average change with trace offset? Yes. It appears that trace offset in Crosslite is the same as a gain change.

Does the magnitude average change with phase offset? Yes, except for Arithmetic Average Magnitude.

In this test I averaged the response of two microphone cables with a second order APF inserted on one of them at 1kHz.

Se deseja saber a média de fontes separadas coerentes ou de uma curva polar, é bom utilizar a aritmética complexa.

Se é média para curvas de equalização, melhor a de magnitude em dB.

Escolhi os tipos que estão mais presentes na maioria dos softwares que possuem funções de média.

Francisco Monteiro

If you want to know the average of separate coherent sources or a polar curve, it’s a good idea to use complex arithmetic.

If it’s an average for equalization, better the magnitude in dB.

[For Crosslite] I chose to offer the types that are most present in most software that have averaging functions.

Francisco Monteiro

L-Acoustics M1

M1 offers a single kind of average, which appears to be a simple average with phase zeroed.

Does the magnitude average change with trace offset? No. There is no trace offset option.

Does the magnitudeaverage change with phase offset? No.

OpenSound Meter v1.0.5

Open Sound Meter offers vector and polar averaging.

Does the magnitude average change with trace offset? Yes, the results are the same for a measured gain change.

Does the magnitude average change with phase offset? Yes for vector. No for polar.

For in-space averaging I use polar method. For vector (complex) you need to have very close phase responses.

Pavel Smokotnin

REW v5.20

REW offers two options for averaging.

Vector average, which averages the currently selected traces taking into account both magnitude and phase. It can only be applied to measurements that have an impulse response.

RMS average, which calculates an rms average of the SPL values of those traces which are selected when the button is pressed. Phase is not taken into account, measurements are treated as incoherent. This does the same as the Average The Responses button. If the measurements were made at different positions (spatial averaging) it may be helpful to first use the Align SPL… feature to remove overall level differences due to different source distances.

REW Help

Does the magnitude average change with trace offset? Yes, but only after the data is permanently changed with the Add offset to data button.

Does the magnitude average change with phase offset? Yes for vector. No for RMS.

RiTA

RiTA offers a single options for averaging traces: Arithmetic Average Complex

Does the magnitude average change with trace offset? Yes.

Does the magnitude average change with phase offset? Yes.

The next version of RiTA will include three options for averaging.

Complex AVG: magnitude estimation is greatly affected when complex averaging is performed. It is useful when you are interested, in close measurements, in knowing the constructive and destructive interference of the sound system.

ABS AVG and dB AVG are intended for spatial averaging of several microphones. Abs AVG tends to give priority to good data and less to data affected by reflections. dB AVG gives equal weight to all data.

By default, RiTA 2.5 uses ABS AVG

Pepe Ferrer

SATlive

SATlive offers a three options for averaging: Create Sum Trace, Complex Add, and Weighted Average.

Does the magnitude average change with trace offset? No

Does the magnitude average change with phase offset? Yes for Complex Add. No for Sum Trace.

SATlive offers 3 different approaches for averaging different measurements.

1. Complex averaging: Will calculate the sum using the amplitude values of each trace and the phase relation between the traces. It is intended to average measurements taken at the same mic position, like Sub/Top time align or interference of different sources. (quick traces -> sum trace complex averaging).

2. Amplitude based averaging: Will calculate the sum by normalizing (center at 0 dB) each trace and afterwards adding the amplitude values only. This is helpful when you want to average traces taken at different mic locations (and in most cases, using the same source).  (quick traces -> sum trace Create Sum Trace).

3. Weighted averaging: This is a special version of 2. where you can assign a weighting factor to each trace (three configurable settings). This was inspired by the Primary/Secondary/Tertiary measurement approach, which I first heard about during my SIM II seminar. In fact, it does not make much sense to add tertiary traces to the result, but it would be possible. (Trace Manager)

Hint: There is a Valid only if all traces valid option for 1. and 2. where you can define wether just one valid result at a certain frequency will be sufficient for a valid result or all traces averaged must contain a valid value to create a valid result.

Which of the options do you recommend to your users for judgement of tonality and EQ operations?

Only option 2. and 3. will make sense here. I rarely work with averaged measurements during EQing. Normally I’d use the Primary Location trace as the base for EQ while the other traces help me to distinguish if the problem is global or just local.

Big differences between the different mic-locations (primary/secondary) indicate a problem that you should fix before applying the eq (redirecting the speaker, additional speaker, speaker with a different directivity pattern).

For overall tonality I’d go for 2 and for Eq-ing for 3

Thomas Neumann

Smaart v8.5.0.2

Smaart offers two options of averaging with the second including built-in proprietary pre-processing.

Decibel spatial averaging, sometimes called arithmetic averaging, is a simple average of decibel magnitudes at each frequency. Spatial power averaging is the average of squared linear magnitudes at each frequency with the result converted to decibels.

Unweighted dB averaging works exactly the same way both transfer function magnitude and spectrum averages. When you select Power averaging for transfer function measurements, however, Smaart automatically adjusts the overall level of all individual measurements going into the average according to their average decibel magnitudes in the range of 225 Hz to 8.8 kHz so that they are all approximately equal in level throughout that range.

Rational Acoustics Smaart v8 User Guide, Release 8.3

Does the magnitude average change with trace offset? No.

Does the magnitude average change with phase offset? No.

Our data is in dB, so we have to decide whether to average linearly or logarithmically, whether to normalize first, whether to weight by coherence (does it make sense that poor-quality data gets as much “say” in the final result as high-quality data?) and of course remembering that FFT math spits out complex data points, not simple integers.

So you can end up with a lot of approaches that are all valid from a mathematical standpoint, but the question becomes “which method gives us the most useful result?” (I could average together the number of socks in my drawer and the number of tires on my car, and even if my math was correct, it’s a meaningless answer for all practical purposes.) So at the end of the day, we want averaging that produces information that’s helpful to the user. If you have a bunch of traces and you average them, we have an expectation of what that final averaged response should look like. How well does it highlight the trends indicated by the individual traces? That’s what we’re looking for when we take an average, and so our averaging is designed with that in mind.

In terms of which to use, just like everywhere else in Smaart: if you’re not sure which setting you need, use the defaults. They’ve been carefully chosen over many years to give good results without the user having to tweak around. I actually reset the software to default configuration every time I use it, and I pretty rarely need to go in and change a bunch of things from that state. The primary advantage of power averaging would be if you’re averaging together a bunch of traces that have severe comb filtering (which hopefully doesn’t happen all that often). The math will give more weight to the peaks and less to the dips, so you end up with something that can be more representative of the overall response in that area and what your ear might tell you. But – in most circumstances, the differences between coherence-weighted dB average and power average end up being very small. If you create both types of average from the same dataset, and lay the two averages traces on top of each other, you’ll see they tend to agree very well. I think you’d have to come up with a pretty contrived situation or have pretty bad-quality measurement data to get a result where the power averaging and the dB averaging disagree.

Michael Lawrence

All together now

Here’s an overview of the different averages being discussed in high contrast. All of these are my own estimations since the math is not exposed and is in some cases proprietary.

Which one should I use?

Please follow the manufacturer guidelines and in most cases stick with the default settings.

The demos in this post average electrical measurements of symmetrical EQ filters in order to clearly expose the calculations being used. I want to be able to see clearly if the average of +6 and -6 is 0 or something else. Measurements of speakers in rooms will feature many wide peaks and narrow valleys instead of this symmetrical behavior.

As I worked through each demo I found myself wondering why I might use one average over another. Being visually inclined and looking at a graph, at first the simple magnitude average made the most sense.

(-6 + 6) / 2 = 0

M1 offers this as its only option and it is the default option in Smaart and SATlive.

Why do the other options exist?

If you had one subwoofer and I gave you another, how much would that be in decibels? You would add 0dB + 0dB to get 6dB.

If I gave you another half a sub you would have 8dB because 20 * log10(1 + 1 + 0.5) = 8.

Following the same process of linear to log conversion, we should calculate the decibel average of -6dB and 6dB like this:

20 * log10((0.5 + 2) / 2) = 1.9dB

Maybe it makes more sense now why some audio analyzers like REW, RiTA, and Open Sound Meter show an average of 1.9dB instead of 0dB.

Interestingly, Bob McCarthy finds even this form of average to be lacking since it does not take psychophysics into account.

Studying summation revealed that 20–40 dB dips are likely to stay down in only a small area, whereas 6 dB peaks may spread over a wide area. Studying perception revealed greater tonal sensitivity to wide peaks over narrow dips. Therefore we should be wary of accepting 0 dB as the best representative here. When samples agree, the averaging builds confidence. When samples differ, the average is suspect. There’s safety in numbers when math averaging is used: get a lot of samples.

McCarthy, Bob. Sound Systems: Design and Optimization: Modern Techniques and Tools for Sound System Design and Alignment (p. 453). Taylor and Francis. Kindle Edition.

In this case “0 dB” refers to the average of a 6dB peak and a -40dB valley.

My takeaway from all of this is that more measurements combined with optical averaging (looking at them all at once) is more important than the specific form of mathematical averaging you choose.

What do you think?

Can you remove reflections from live measurements for more accurate alignments?

By

In my last article I found some minor success in validating subwoofer alignment indoors through the use of carefully chosen microphone locations and their average. Next up: Can it be done live in real time?

Thierry De Coninck sent me an article he had written outlining a procedure that would allow you to not only create these averages, but also set the delay locator for your transfer function based on distance to avoid any errors from reflections or lack of HF content arriving at ground plane. I also found an AES paper called Measurement of Low Frequencies in Rooms, which outlines a similar method through the post processing of IR measurements.

A new method is outlined for measuring frequency response at low frequencies. This method uses microphones arranged in a low frequency end-fire array to create useful directivity to discriminate against sound waves from rear wall reflections and reverberation. It also operates in the time domain, processing the acoustic impulse response as it arrives at successive microphones – a shotgun microphone writ large.

D. Murphy, “Measurement of Low Frequencies in Rooms,” Paper 9482, (2015 October.).

What interested me most about Coninck’s method, though, was the ability to get results in real time. Here’s the outline:

  1. Deploy two microphones (or more, potentially).
  2. Measure the furthest position setting its delay locator automatically or by distance from the source.
  3. Set the delay locator of the next measurement position manually based on the difference in distance to the source from the previous measurement. (eg. If the second measurement position is 1m closer to the source then offset the delay locator by 1m / speed of sound.)
  4. Verify that the LF phase of both measurements has some agreement.
  5. Create a live average.

Initially I thought I would create a script in MATLAB to automatically calculate the delay locator setting for each transfer function pair using pythagorean’s theorem, but looking at some initial results convinced me that, if possible, setting the delay locator automatically from the IR would be more accurate. Since my goal is to measure and align at head-height then the microphones should receive enough high frequency content to make this possible.

The Setup

Here’s a section view of the sound system at the Sojourn Campus Church here in Minneapolis. Adam Rollin from AVE was kind enough to let me tag along and do some tests during his normal system calibration work.

I put these details into the Sub Align calculator and could see we had a small phase span of 28º. I decided we would do our alignment at the last row and distribute the microphones evenly from there.

Here’s a comparison of all six measurement mics at about mid-depth of the audience after level and polarity match (1/24oct smoothing). You can see that we’ll have our work cut out for us in the low-mids.

Alignment

We used the Relative/Absolute Method to complete the main+sub crossover alignment. Here are the native solo measurements at 1m. The sub trace is offset -6 dB due to half-space loading.

At first I thought the wonky phase around 1kHz was there because the mic was too close, but we found the same thing in the far-field. I’ll have to save the story of that polarity inversion for another time, though.

I wanted to reduce the overlap and improve the alignment without adding a lot of delay. An LR48 HPF was inserted on the main at 42Hz. This left us with the pre-alignment value of 2.4ms of delay on the main and a polarity inversion.

Why the polarity inversion? Compared to other possible variations it allowed for less delay and improved alignment.

Here’s how we deployed the microphones.

Here are all six traces with the average of the main.

At first I though, “Success! We removed the reflections.” Then I noticed the wrap at 200Hz. What the hell?! You’re supposed to disappear!

I can tell it’s not supposed to be there by comparing the average with the near-field measurement.

Where’s it coming from? Is it a floor bounce, a ceiling reflection, or something else?

If it’s a floor bounce, shouldn’t we be seeing a 2.5ms peak in the live IR graph in every measurement? Not necessarily since any peak in the low-mids would be down ~40dB and hardly visible.

There’s one way to rule out the floor-bounce for sure. Here are the ground-plane measurements.

Here is the average ground-plane vs near-field. Good-bye 200 Hz phase wrap.

Why does this matter?

The wrap at 200Hz caused by the floor-bounce causes phase shift through the crossover region and therefore misalignment.

What’s the solution?

Don’t align to a reflection. Do align to direct sound by using the Relative/Absolute Method or SubAligner.

A distance measurement from the alignment position were almost matched so we decided to leave the pre-alignment delay value unchanged. Here’s the alignment at head-height.

Everything is within the corridor of 60º, but if you wanted to be picky you might decide to use less delay in the main to attempt to close the 30º phase offset at 70 Hz. This may create a misalignment, though, because the floor bounce is creating the extra phase shift.

Attempts to remove the phase wrap at 200Hz by varying the microphone positions over height and width were unsuccessful. A majority of the traces always included a dip around 200 Hz.

Conclusion

I started down this path a year ago in order to prove the results of the Relative/Absolute Method, and therefore, the necessity for a tool like SubAligner. The Catch22 is that the more you need SubAligner, the harder its results are to validate. If the phase trace is inactionable, SubAligner can’t fix that. It just ignores it. And while a measurement at ground-plane will help with visualizing the data, it will hurt the alignment because we don’t have ears on our feet.

Using an audio analyzer to validate SubAligner’s results is like trying to validate the presence of an optical illusion by taking a photo of it. It will only enhance the illusion.

Here are some takeaways from Sojourn:

  1. No amount of measurement data and mic positions can guarantee a reflection-free average in the far-field (especially with poor direct-to-reverberant ratios). The null created by a floor bounce or other boundary may be sufficiently common to each individual trace that it ends up in the average.
  2. Since a measurement cannot be guarantee 100% reflection free, there is always risk for misalignment. Gathering more data in a strategic manner can help reduce this risk, but the only real solution is to abandon the audio analyzer for a more appropriate tool.

Audio Analyzers: Pink Noise vs Sine Sweep

By

Why do some audio analyzers default to pink noise and others to a sine sweep?

They are different acquisition methods, but they’ll both get you there.

Real-time

  • 2 channel input
  • Measure in real time with any source material
  • Common among live sound engineers

Non-real-time

  • 1 channel input
  • Measure off line with a pre-determined source
  • Measure THD
  • Get very high SNR
  • Characterize decay (T60, C10)
  • Common among acousticians, manufacturers, mobile electronics technicians

You don’t want to feed a real-time measurement with a sine sweep because the real-time analyzer wants a broad-band source, which makes the sine sweep look like mostly nothing. It’s like putting unleaded in a diesel engine.

Transcription

The transcription was automatically generated. Please let me know if you find any errors.

This year, I’ve been getting a lot of questions about why some audio analyzers use pink noise and why some use sine sweeps, and it started around the time that acoustics came out with their M1 P1 platform. And so I think that was kind of what people are looking at is like, hey, wait, we’re used to kind of hearing pink noise when we use things like smart. Why is now the acoustic system using this sweep?

Should I be using a sweep? What’s going on here? And I can go directly to them and maybe I’ll try to make another video later where I talk to Vic or Scott or someone from acoustics and just chat with them about why they made that decision. But I thought it’d be fun to talk to you about just make a short video about just some general places where we see the show up so that people can just including myself, can have a better idea of like maybe why and why some people prefer it and why some platforms prefer it.

So this is the first question. Why does some audio analyzers use pink noise? Isn’t some use a sign sweep?

Yeah, and that’s really the great place to start. And I think, you know, what we’re really talking about here is not which signal should I use?

But what we’re talking about under the hood is what mathematical method is this analyzer using to get the data right. And that’s something that as an end user we don’t care about. We don’t want to get under the hood and know what math it’s using, but that we’re talking about two different types of math that will give us the same answers. So one type of analyzer is designed so that it can use any signal that we want, and that’ll be the dual channel fee.

So it’s source, independent. Right. And then the other type of math works by using a very specific, mathematically predetermined signal.

So a sweep would be a good example of that.

So there they’re just different acquisition methods. And I think as we’re going to see as we go on here today, you know, they’re both going to get you there.

They’re both going to give you that answer.

If you’re just trying to say, hey, you know, I want to see the magnitude and phase of of this loudspeaker, whatever they can both produce that they just use different mathematical methods under the hood to produce that information. Got it.

OK, so would it be fair to call these should we break it down? Is it real time and not in real time or is it real time? And I are mode. What do you think is the better way to distinguish that?

Well, so the real time measurement that is the source independent one. And it’s happening. It’s happening in real time. Like you said, I can be tweaking a delay or it can be tweaked and I’m going to see those results to show up the measurements continuously running. And that’s the one that can use anything that we want for for a source material. Yeah, non real time would, I guess, be a good catchall term for something like a sweep where you say, OK, run it and it goes and then the computer goes and kind of thinks about it.

And then it brings you your answer. Right. And then if you want to tweak that IQ or something, you’ve got to go run that measurement process again. So it’s it’s not happening in real time. In Smart specifically, we call our non real time mode impulse response mode because we’re gathering the impulse response system. But as you know, once we have the impulse response, we can transform that and get get the frequency response. And either way.

So so that’s our name for it. But yeah, I mean, we’re fundamentally distinguishing between measurement that’s happening continuously in real time and a measurement that’s happening once and then it’s sort of a one shot thing. Yeah.

And just to give you an overview of, you know, there are a lot of these audio analyzers like Smart that I think can do both modes. But just to give an overview of some where we where we’re used to seeing them, they are things like when it comes to the real time mode, we’re used to thinking of things like Smart sat live and now we have open sound meter. That is there’s probably some that I’m missing. But then when it comes to the non real time, I immediately think of like RW for our capture, wave, capture or wave tool.

I can’t remember the name. Is there any that I’m missing that you think are common? Oh, I mean, there’s tons of them.

I think another common real time dual channel platform would be Sistan that people are familiar with. There’s a ton of sweat measurement platforms out there. Generally, it’s it’s a sweep or some other period since signal versus the real time dual channel platform. So in terms of the mechanics, the whole point of the real time dual channel platform, like in Smart, like you’re setting up to inputs the measurement in the reference. So we’re just saying, all right, let’s how do these compare?

We can continuously run that comparison.

Whereas when you’re doing something like Remix Wizard, you don’t have the necessity to do that reference channel that loop back. Right. It’s just the computer is going to produce that signal. It’s going to come out and then it’s just going to record it comes back and you could record that sweep to us CD and play it in a different building and record it.

Just bring it back and. So you have are you inputting one signal or are you inputting a pair of signals? Right. And so that’s the fundamental difference here, is that when the dual channel real time, that the whole point of that reference signal is where we don’t need any particular predetermined signal to do this, just give us a copy of it versus the suite measurements that that that sweep is very particularly created by the computer. And so when it comes back in the computer, you can compare that suite to what it knows that it generated and that’s how it produces the measurement data.

Cool.

And you and rest on Acoustics recently published a great video that I believe is called Pink Noise versus Sweep. Or is that was the title of that one?

Yeah, I think well, it was kind of had a double title, right, because it’s exactly what we’re talking about. A lot of people say I think it’s called like in parentheses or in quotes. It says pink is sweet and then it says real time versus real time versus impulse response. Right.

So people people should watch that one for sure. And I just want that one yesterday. And it’s the same topic. And you do some great demos in there. But I know you have another demo. So before we dive into this demo that you’re going to show us, let’s just talk about some common places where we see these two things show up. So starting with real time mode, a lot of people, especially those of you like you and me and the people watching this video, we are used to seeing most commonly this real time mode using things like smart when we are doing temporary installs, a lot of concert, sound theater, that kind of thing.

Do you think that’s right? And where else do we see real time mode show up?

Yeah, I would I think it’s a fair statement to say that the majority of people doing live sound system optimization work are probably using a real time tool. And I think the reasons for that are obvious. The fact that it happens in real time and the fact that we can use any test signal, we want those to have some real obvious benefits in a production environment. Obviously, a lot of this is about how quickly we can move. And so there are some real benefits there.

Now, you can obviously do this type of work with this measurement platform. I know I have a couple of good friends who used Rumiko. Was it for a long period of time to do this work? And there’s nothing wrong with that. And I want to stress that, that this is all about you using the tool that you’re comfortable with it give you the results that you need. But given the choice, typically I’m trying to move as quick as as quickly as I can.

I’m going to go for a real time measurement platform for for a lot of this type of work where it’s, hey, we want to just measure the response to this sound system so we can put a couple filters in or I need to dial in some delay times or stuff like that. I just want to continue measuring during the show.

Like, you never stop you potentially. Yeah.

Yeah, exactly. I mean, obviously you can’t do sweeps during a show. Right. So so there are there are times when the circumstances in which we’re trying to work are going to dictate a certain workflow, and that would certainly be one of them. Now, if we want to talk about the sweat measurement platforms or what we’ll just call non real time measurement platforms, the really big benefit there is that the analyzer is able to sort of cut through a lot of the sources of noise that come into our measurements.

So we have a couple of different sources of noise in our measurements. Want to just acoustic noise or at the back or my furnace is running right now.

So that would show up my bird screaming in the next room, all of that stuff. Right.

So so that is something that we deal with both in real time and on real time measurements with averaging the more average that pushes that noise down. But there are other sources of noise and sometimes we want to deal with those in our measurement.

So one source of noise comes from the way that the analyzer has to sort of chop up. When we’re talking about real time measurement, we can use anything that we want. It could be music, whatever. We’ve got to feed the analyzer specifically sized chunks of that signal in order for it to do its thing. And so do the process that is involved in that can contribute some noise. So that’s when you get into a period matched stuff where we’re using a signal that’s the same length as the analysers time window.

So it’s smart. That’s pseudo random noise, which we can we can demo or the sweep. The sweep is going to be you set this wavelength in the analyzer. So that eliminates that source of noise and it can bring down the noise, floor the measurement. And then then finally you have non-linearity of the system. Loudspeakers have harmonic distortion. Right. And so when we have all the other sources of noise in our measurement that’s so down low that we don’t really care about it.

But when you’ve stripped back all the other layers, if you’re doing acoustics work right, this is really where the sweet measurement really comes into play, is if it’s your job to do acoustic treatment for a venue. And you really need to characterize that. Talk about the 60, right, I want to measure 60 dB of clean, reverberant decay, I need a really good signal to noise ratio in my measurement. So now we are stripping back those layers of noise and using you know, you’re talking about I’m going to use a really long sweep.

I’m going to do eight averages. So you’re talking about a 30 second acquisition time? Maybe, but you can get down 60, 70 dB into that reverberant and you can get really accurate characterizations of the reverberant qualities of the space. If you’re a contractor, that might be something that you really need to know. So so that is a situation where I’m willing to spend an extra 30 seconds to get this data and get really, really, really good clean measurement data way down on the noise floor.

And that’s that’s a situation where I definitely go to impulse response mode and smart.

And I’m going to I’m going to spend that extra time and I’m going to get these really clean impulse responses. So like you said, it all comes down to what what are we trying to do when we pull it analyser out? And this is something that I know you’ve talked a lot about in your videos. Right. What is the question we’re trying to answer here? What are we seeking to learn? And so if if it’s I just want to enter the system real quick and put some filters in or whatever.

I don’t need 70 dB of signal to noise ratio because that stuff lives up up at the top. And I’m going to get my answer quickly. And that’s great. And if I’m talking to a client about how much acoustic treatment should we order, I want I want to see that.

I want to know what’s going on. 40, 50, 60, dB down. So so that’s kind of different. Different tools for different jobs.

I remember a couple more jobs that we mentioned the last time you and I talked as well. You talked about a manufacturing lab and I was like, oh yeah, they have these specific conditions they would want to know in real time zone could really benefit them. And they want to see harmonic distortion and they’re looking like really close that drivers and they want to get really deep. And so that’s a common one, right?

Yeah, a lot of people I have some friends that work in loudspeaker design and that type of thing, and they’re typically working in way, way more controlled conditions than than we typically. What did it get? Right. So they’re in probably anechoic chamber or really, well, acoustically treated environment. And they are also measuring for a much higher level of of resolution, I’ll say, than than we typically would, because we know that in a venue we move our mic over a foot and we’re going to see a different answer.

Right.

When you’re designing a loudspeaker, they really need to see sub dB resolution on a lot of this stuff. And so so, yeah, they’re going to use typically a sweet measurement in a lab and they don’t care about the acquisition time. They’re not at a show. They don’t have a soundtrack coming up. Right. So so they can take longer acquisition time measurements. And the benefit of the sweet measurement is, like you said, because we’re only putting one frequency at a time through the system, we can keep track of what’s coming out and so we can spot that distortion and we can actually separate it out with a sweet measurement.

So if you’ve worked with Rikyu, is there there’s there’s an ability in there to show harmonic distortion over frequency and each each harmonic, whereas in a real time measurement, the distortion is just part of the tonal response of the system. If you’ve looked at I know you’ve done a little work with showing your viewers the noise test and we see what happens when you get distortion in there. The coherence starts to come down. Right. But in a sweet measurement, we can actually separate that distortion out.

And instead of saying it’s part of the tonality of the measurement, it’s part of the tonality at the system, we can actually split that out of the way. And the benefit there to go back to the acoustics work is if you’re trying to measure the acoustics of a space and you accidentally drive the loud speaker into distortion when you’re measuring the space, it’s separated.

It’s OK. So the ability to separate out distortion is maybe something we may or may not want in a real time situation. But it’s definitely something that’s really important in a controlled condition. If we’re studying distortion, I do work with the rental company here in town and after a month of shows, I’ll take all the boxes and I’ll sweep them and I’ll compare those measurements from what they were a month ago. And that’s a really, really great way to spot a blown driver or another problem with a box that needs to be repaired, stuff that you might not hear with just listening to music or listening to pink noise at a restaurant.

It’s like a health checkup. Exactly.

And so so, again, it’s all about that. We have different ways of taking these measurements, depending on what question we’re trying to answer. One thing I was going to add is that last month I did a training for a bunch of mobile electronics technicians, people who are installing high end systems in cars and trucks, and we decided to use Rumiko Wizzard partly because that’s what they were already using and they were familiar with it. But it turned out to be super useful because they don’t have their client standing next to them, waiting for them to get done in a few seconds like they’re working by themselves in the shop.

And if they needed to if the deadline was tomorrow, they could work all night if they needed to. But my point is that it was really useful to be able to take a measurement and then do a lot of things with it offline and has some great abilities to do, like filter modeling and things like that. And it was great to see that on screen before you take action. And we also mentioned being able to do these offline measurements. So it might be difficult in something like a vehicle to like get your impulse and get your signal generator into the input of that thing.

And so you could just burn a CD. If the car is a CD player with the signal that you need, play it back and record it and do your work that way as well.

Yeah, exactly. So for all of the reasons in a in a busy production environment, why a real time measurement is really great. There are some situations where it just is more complicated to do that, like you said. So if you have a friend who works for else. So how do you measure a smart speaker? Right. That’s all controlled control over wi fi and Bluetooth and stuff. You can’t really get an input into there in real time.

So that’s a perfect example of a situation where you’re like, no, I have this prerecorded test signal. I’m going to just play it back through this thing, pull it back to the analyzer and do what I need to do. And so, yeah, sometimes a non real time measurement is is really helpful. Cool.

So can we take a look at your demo and then we’ll talk about how screwed things up?

Yeah, absolutely, man. All right, let me I’m going to share my my screen here. Can we see this?

I see it. Great. All right. So so this is familiar hopefully to a lot of your viewers. We’re just we’re here and smart. We’re in the default transfer function view. I will point out I am using the version eight point five beta. This beta is public. So if you have a smart version IT license, you can go download this right now. And version eight point five stable will be out in a couple of months. And there’s a reason I’ve chosen this to show you this demo on me because we’ve completely overhauled impulse response mode and it’s really significant and it allows me to look at multiple impulse responses at a time, much like we can in real time mode.

So that’s really, really useful for for this demo in particular. We’re going to be comparing stuff. Right. So I’ll start off.

I’ve got a there’s a loudspeaker and a microphone over here in the corner of my office. And I will just start off by just gathering. I’m going to run the real time measurement for a second. Just take the transfer function here. So apologies for the for the noise. And what do I need, my speaker went to sleep on me. All right, now that it’s woken up, OK, so there’s my transfer function measurement. I’m just going to save this.

Great. So, you know, as you see, as soon as I turn it on, as soon as the system comes up, it starts doing its thing right. And so it’s just going to kind of continue on. I am going to activate the setting that we have, which is to stop the generator once I capture the measurement just so we don’t have to listen to the noise. Oh, cool stuff.

Right in the new feature. So. All right.

So we’re going to flip over to impulse response mode. And the way you do that in Smart as either press the ickey on your keyboard or you just have this little impulse button down here.

And you’ll notice that now this looks a lot more like real time mode than it used to, right? I have my data bar and all that stuff, so this is pretty cool. And I’m going to delete some old data so we can have our fresh new folder notice over here on the right.

Same transfer function. Engine looks exactly the same. I got my two single monitors, I got my delayed time, all that stuff. So that’s the same deal. So the three parameters we’re going to look at here are changing the 50 size. Using a longer field, give us a better signal to noise ratio, adding some averages and both of these things, if you’re familiar with Mickey Wizard, these are both parameters that you can control.

You can say, I want to use a longer sweep and I want to use more averages of that sweep. And then the third parameter we’ll look at is what happens when we change the signal type. So let’s start off with the default settings. We’re going to use the same random pink noise that we had before. So this could be music, could be whatever you want it right. It’s still a dual channel measurement. We’re going to use the 16 KFT and no averages.

And I’m just going to fire this thing up and let it go.

So let’s say that is a random 16 zero averages, right? So this is this is a basic it’s a really notice to talk about a third of a second to come through. So that’s three point forty one milliseconds.

So let’s see if we can lower the noise floor here. Let’s try to get a bigger dynamic range. Right. So one thing we can do is use marriage. So let’s go to let’s go to eight averages.

All right. And we go. So now you can see that it’s lowered that that noise floor by a good, what, 12 dB or so. So that’s pretty good. That’s pretty cool. So let’s now go to a different type of signal. So what we’ve been using up to this point is just random pink noise. Like I said, it could be music, could be whatever you wanted.

If we use pseudo random pink noise, what that does is it’s going to match the length of the pink noise signal. It’s going to be repeating signal now. And you can actually hear it repeating. It sounds like like an old Atari game or something. And it’s going to automatically match to the size of the fee they’re using. So when I hit this button that’s going to drop out, that processing that I was talking about, it’s called a Daito and it basically spoonfeed the analyzer, the correct chunk, so to speak.

But if the signal that we’re generating is already the right size, we don’t have to do that. And that eliminates the source of noise in the measurement.

So before you hit play, I just wanted to I think I’m noticing like this is kind of the classic thing you might think of, like popping a balloon and just recording it in a room right until now.

Yeah. So so we’re doing a dual channel version of that. Right. But yes, you can absolutely. If I if I hit this record button here that turns this into a single channel, direct air. So yeah. Then I could hit play pop the balloon or clap and I’m going to see that it’s a very similar thing. Obviously there are some advantages to the dual channel approach and that we can see the popping and clapping is really tough, right.

Because it’s so short, you’re not putting a lot of energy into that room.

So something like a sweep or pink noise test signal, since we can excite the system for a longer period of time, you don’t have to be so loud in the space. It’s easier to get a good level on your measurement and get above that noise floor. Whereas for for a pop or a clap, if you’re playing that through a you know, you’re probably going to have to push that system right to its limits just to get to get enough energy.

So so there are some benefits in doing this, not as a direct impulse. And of course, we are creating the same result. Right. We’re still creating that impulse response here. We’re just doing it with the dual channel method. Does that make sense? Yeah. Thank you. Cool. All right.

So let’s do the same thing. The only thing we set this time is instead of using random pick noise, we’re going to use the super pink noise and I click the drop our data window and that’s going to sink the pink noise signal length to the size of the fees. I’m going to fire that up.

Nobody can hear it over the zoom audio, but it sounds a little bit different. So that’s that’s a pretty big improvement, right? So pseudo random, let’s call it 16 averages. So what we can see is the noise floor is pretty drastically lowered, right, by a good 18 dB or so. And now we can see that the reverberant decay of the room is starting to be exposed.

That’s cool. Remind me, why do we want to lower the noise floor? I think maybe we covered this at the beginning, but at this point I’ve forgotten.

Well, we may we may or may not want to. Right. So if we’re in, we’re going to finish up by comparing this the frequency responses of these to the real time measurement. And we’ll see that they’re going to be pretty much the same. So, again, if you’re just measuring because you want to kupa, you don’t need to bother with lowering the noise floor because that stuff happens way up on the top. You know, our world was the top 10 dB when we’re we’re not doing stuff that’s 70 dB down.

But if we’re acousticians and we want to characterize the reverberance of a space this up here, if I’m hitting the noise floor only only 30 dB down, that doesn’t give me a whole lot to go on. Right. Whereas if I can push that noise floor down, I’m exposing more of the reverberant qualities of the space and I can do that more accurately. So for an example, I can click the T 60 button and I can you know, you see these markers come into play and they don’t really land quite like they’re supposed to.

We’re generating all these acoustic metrics if we’re for an integrator or we’re trying to study the space itself. And what you can see is as we lower that noise floor, you’re going to see these metrics get more accurate and that marker placement will improve. So so the software just can get a cleaner data set to calculate those types of things, basically. Thank you. Yeah, absolutely. So let’s nuts. Let’s now go to a longer fee. OK, so let’s go up to 60 for KFT.

So that’s a little over a second long at forty eight K sample rate. And so we’re talking about one point three seconds it averages. So now we’re talking about nine or 10 seconds to get to get a measurement, but you’re going to see it’s going to lower that, that noise floor even further. All right, so this looks way different than the other ones, right, so it’s longer. Like we said, you can see that it extends this is this the axis down here?

The X axis is in milliseconds. So we can see over here one point three, four K. It’s not a frequency. That’s that’s a time. Right. So we’re over a second long compared to these guys, which are about a third of a second long. But you can also see that, again, that noise floor is significantly lower than it has been.

So we’ve got even more of that reverberant decay exposed. So I’m going to save this. This was a sixty four K at fifty and we did eight averages. And so in in smaller version eight point four, you could do this work, but you can only look at one trace at a time. And so that’s why I’m using eight point five for this demo, is that it’s obviously a little easier to see this stuff when you can overlay them. All right.

So now let’s go for the sweep. Right. Let’s let’s go for for the big guns in the sweep has that additional ability. So we’ve done averaging. We’ve lowered the acoustic noise. We’ve kind of dealt with that. Then we used a period match noise and that allows us to get the data window out of the picture and that that brought the noise down more. And so when we’re this far down, we’re talking about, you know, you can see where we’re below 70 dB down now, pretty far down now.

We can start talking about. But what about the distortion products of the loudspeaker? And so when we go to the sweet measurement, instead of those being mixed into the noise of the measurement, it’s going to separate them out. And so we’re going to see that. So I’m going to pick a little bit of a longer time record because that works a little bit well, better with a sweet measurement. We’re going to select Pink Sweep as our material here.

And notice I’ve got a setting. One thing I’m going to do is turn this down just a little bit because we don’t need it to be as loud the the signal gen level and smart. This is actually a good thing to remember. This is a peak level. So we’re using pink noise simply as a crass factor, about 12 dB.

So the average the arms level is going to be you below the value in this box when we go to a sine wave sine wave as a crass factor of three dB. So that’s going to sound a lot louder even though the level setting looks the same.

Right. And so it’s just it’s I’m going to just pull it back a little bit.

So the box that I’ve checked here is triggered by impulse response. And so when we get into sweep’s, it’s really important that the sweep lines up with the data window of the analyzer or the time record of the analyzer. Excuse me. So we want that the analysers time record to start. And then we wanted to hear the sweep and then we want to leave enough at the end that the reverb in the room can ring out. And then we want to stop listening and then we want to process that.

And so manually synchronizing that, you can really get into some trouble with getting goofy results because you had timing problems. So that setting when I do this, it’s automatically going to trigger the suite for me when I hit play, which is really cool. I don’t have to worry about trying to sync it up. Right. So we can do a three second sweep, two point seven seconds, and I’ve chosen eight averages. So we’re going to watch this thing go for a good thirty seconds, but you’re going to see the additional benefit of doing that.

OK, and know what, let’s let’s not go crazy. Let’s go to four averages here just because people’s time is valuable these days. All right, let’s let’s hit it. Let’s see what it does. Maurice. There we have it, so we’ve lowered that noise floor even further. And so now we’ve got I mean, I’m at I’m at negative dB here, so we’ve really exceeded our original measurement.

The pink the pink one here, it was minus 30 or so, minus thirty three.

So we’ve gotten pretty much a good forty five or fifty additional dB of dynamic range out of this measurement. So I’m going to save this one to make for averages. And it’s a little bit it’s not super easy to see, but if you have a lot of food in your system, these peaks that you’re seeing here at the end. They’re not super clear on this measurement, but those are the separated out impulse responses of the additional distortion products. So you get an IRR for your fundamental, which was the sweet tone that we hear, and then you get a separate IRP for your second harmonic.

And it is it’s pretty neat.

Now, again, on a super good linear loudspeaker, this doesn’t really get you all that much because you don’t have that much distortion in your measurement to begin with.

But if it’s a if it’s a speaker that’s working really hard toward it, toward the limits of what it can do, you have high distortion in there. This can get you like we see we gained an extra 60 or so by doing this. And so now I can come in and I can hit my 60 button. You see, now those markers really fall along the deck where they need to. And I can come in here and I can look at if I’m going acoustician.

All this stuff means it tells me a lot about the energy arrival’s in the space and reflections and the tonality of the reverb and all that stuff.

And so there’s a lot of stuff that that when you and I walk into a space and we’ve got to mix to show we’re just like MAPP, you know, we this doesn’t really do anything for us because we can’t do anything about it. We’ve got to deal with it. We’ve got the best we can do is not hit the walls with our speakers. Right.

But if your job is to work on how sound decays in a space, if that is your living, this stuff is really important to you. Right. So so we’ve really been able to do a good job of characterizing that stuff. Now, I am going to go to two pane view here and you’ve got linnear.

You can actually do some really cool stuff here. We can we can actually render a spectrograph of this stuff and you can look at histograms so you can see 60 in octave bands or third octave bands. And there’s all kinds of cool stuff you can do in this mode. But I’m going to go to frequency. And so we’re we’re seeing basically the magnitude response of that data. And I’m going to apply a little smoothing just to get rid of the of the grass here.

OK, so what we can see is the Pinki, which was really, really short. He had a pretty lousy signal to noise ratio.

You’ll definitely see some wacky stuff happening in the low frequencies down here where my speaker is not putting out a ton of energy.

It’s a little speaker. So so the signal noise ratio down there is not great and we see that. But if you look at the overall trend here, these are all telling us pretty much the same story.

You can see that if our goal is I want to know the frequency response to this loud speaker, we pretty much got that. And what I’ll do is if you just sort of take note, hold this image in your mind and I’ll send you a graphic of this so you can kind of do a side by side in the video. If you want to make it easier for people to see.

I’m going to go back to real time mode. And I’m going to pull that measurement back up, and you see, again, if I put the same smoothing on, we have the same story, right?

There was a different. OK, so it’s a different vertical scale in our mode. But, you know, what you can see is this measurement happened instantly and gave us the same answer. Now, how do we know there’s noise, that there’s not noise in this matter? Well, that’s what the coherent trace does.

So you can see down here where we have the wacky behavior in in the the impulse or field measurements, we have low coherence. And that’s telling us the same thing. Hey, you know, there’s some some funky stuff going on here with your data. There’s noise. We’re not it’s not super high quality data. So, you know, I think we return to the first thing that you said, which is what am I trying to learn if I just need to measure system so I can IQ it?

I have no problem using a real time measurement for that, because it’s given me the answer that I need and I don’t have to wait. And like you said, I can tweak my IQ in real time or whatever, and I can just watch that. And I don’t have to go through that process again where I go take a bunch of sweeps.

So so, you know, going back to to this stuff, if if your interest is characterizing the decay of a of an environment, then yeah. Then it then it definitely makes sense to get into these specialty types of impulse response measurements or we can use specific signals and we can really dig down into that noise floor.

And that’s a real benefit, you know. But, you know, they’re there. I wouldn’t say that there’s one type of measurement that’s inherently superior. I think that they’re they’re different tools that are most effective at answering different types of questions.

Awesome, yeah, that’s that’s a lot more clear to me now after seeing that demo. Thanks, Michael. Yeah, absolutely, man. He did ever tell you about the time that I borrowed my friend Mark’s diesel Volkswagen Beetle? No, I would love to hear the story, though.

It’s actually pretty sad, but I do things like this in my life all the time. So this is back when I was living in Texas for a little bit after I’d moved back from living in Slovakia. I didn’t have a car and I was living with my parents for a little bit. And so I needed to go somewhere. I think it was actually I had a gig and I borrowed my friend Mark’s car. And then I was driving back home and I was like, Oh, I’ll do the nice friend thing and I’ll fill it up with gas.

And I did. And then I got back home and then I was driving it back to his house from my house. And then it stopped in my driveway and it wouldn’t turn back on. I was doing all the things that he taught me, like, you turn the key and you wait for it to the light. Come on. And then you started up and I could not figure it out. And I sat there for a long time. And I think it took me like 30 minutes before I was like back at the gas station.

It says all over the car, diesel, diesel, diesel, diesel. I just never you know, I don’t drive a diesel. So I of course, I pulled up and fill it up with I’m going to guess. So if you ever wonder how much it cost to fix that mistake and it cost one hundred fifty dollars to get the tank drained and to then put diesel gas in there again. So recently I posted a video where I was demoing using ping sweep with the transfer function and I posted it.

I was like, cool, this will be helpful. And then Christine just called me from National because they said, hey, you can’t do that. And I was like, What do you mean you can’t do that? Look, you can do this. You just go to a transfer function and you set the signal and he’s like, but you shouldn’t do that because that’s not what it once. And I was like, what do you mean? So so, Michael, why can’t I why shouldn’t I use the pink sweep when I’m just trying to do a transfer function?

Right.

So in real time mode, you know, there’s a reason that we use pink noise even though our measurement system is independent. Like, you know, we always say, yeah, you can use the feed from the soundboard, you can use music that you like. The reason that we tend to use pink noise anyway is that pink noise has all the frequencies at once. And so the measurement is acquired really quickly. So you can you can take that real time transfer function measurement with your favorite song, if you like.

You’re just going to be waiting a little longer for those little gaps to fill in. Right.

So if you think about the fact that when we’ve got holes in our spectrum, you know, we’ve got to wait for the analyzer to be fed some information at that frequency before it can show us the result.

And so when you use a a sweep, you’re talking about a signal that only has one frequency in it at a given time.

So all of the analysers bins, except for one hour waiting around going, well, I didn’t get anything yet. Right. And then when you combine that with a modern dual channel analyzer, has the multi time window or something similar where those time records up at the top end of that frequency response, they’re really, really quick. So they may just not get anything entirely.

So if you go into again, I don’t recommend that you do this, but if you go into your transfer function options and you choose something other than multi time window, you choose a 60 50.

So we’re generating that whole frequency range from one foot size.

Then you will see some data populate there with with the pink sweep. Now, it’s still it’s still not a great way to acquire the measurement because, you know, you’re talking about we have we’re going to run run more averages if we want to stabilize our measurement. You’re still feeding in a signal that is mostly empty and so for all the reasons that the MAPP that the impulse response measurements really like that right there, they’re waiting. They’re going to get the whole thing and then they’re going to go and process it and give you the result we’re trying to do in real time.

We’re trying to go as fast as we can. And so when you’re feeding in a signal that is mostly nothing, it’s obviously not the best way to do that. So much like you have the diesel car, you just put the wrong type of gas in it. Right. So we’re talking about two different types of measurement engines here. And you just you know, they can’t drink each other’s fuel. It’s the same thing. But as we saw, you know, they’re both going to get you where you need to go.

If you just want to know the frequency response rate, it’s just about feeding the mass what it’s expecting. And likewise, if you go into Roomi, Q Ezzard and you feed it some pink noise, that’s not going to work too well either. Right. So it’s all about the analyzer being fed, what it’s designed to, to process.

Yeah, that’s a really good point because in Rumiko as if you’re playing the signal that’s designed to reject all of that other noise and then I guess the same thing in Smart you, it knows what signal to expect. That’s why we have the loop back. Yeah. So, so this is all making more sense to me. Michael, thank you for, for coming on to talk to me about pink noise versus science. Is there anything just kind of on this or what do you want to say to kind of wrap up?

I know there’s like a take.

We talked about a lot of things here, but what’s kind of the take away that you want people to have might take away would be that you should use the measurement platform and measurement approach that that works best for you.

I think things like this where you gain an understanding of why these people choose to use this method and why these people choose this method. You know, I’m not I can’t speak for any manufacturer in why they chose to go one way or the other. But when you know what these measurements can do, then you can decide, well, OK, for the stuff that I’m doing, I think this this is going to be a good choice for me. As we’ve seen, they’re both going to get you there.

It’s all about at the end of the day, we want to just get our answer. We’re going to do a good job. We want to get a call back. Right. We want to we want to keep our gig. So I would encourage people to choose a tool that they’re comfortable with and that they understand and that they can get good results with. And as long as you’re doing that, I think that’s the way to go.

1 Graph Setting You Need to Change in Smaart for Faster EQ Decisions

By

Subscribe on iTunes, SoundCloud, Google Play or Stitcher.

Support Sound Design Live on Patreon.

In this episode of Sound Design Live I talk with the founder of SIA Acoustics and SIA Software and the originator of Smaart©, Sam Berkow. We discuss acoustics, sound system design, and audio analyzer pet peeves.

I ask:

  • How did you get your first job in audio?
  • What’s one of the best decisions you made to get more of the work that you really love?
  • You have managed to build a business that successfully marries acoustic consultancy and system design and integration. It seems like these two jobs would always go hand in hand, but they don’t. Is that because sound system design is a much younger field? Could you talk about what separates and joins the two?
  • What are some of the biggest mistakes you see people making who are new to audio analyzers?
  • Is it cheaper to make a room quieter or make the sound system louder?
  • Tell us about the biggest or maybe most painful mistake you’ve made on the job and how you recovered.
  • From FB
    • Kip Conner: What happened to the Tacoma Dome case study? Can it be reposted?
    • Jason Kleiman: Does he have any advice and/or opinions on using FIR filters in system design and optimization. What is an example use case?
    • Cuauhtémoc Méndez: What are his thoughts on “immersive installations” and their future. Will it last?
    • Aleš Dravinec: Ask him how Kayden is doing.
  • What’s in your work bag?

It’s always cheaper to design it right the first time.

Sam Berkow

Notes

  1. All music in this episode by Wowa.
  2. System toning songs: Ali Farka Toure & Toumani Diabate – Debe, Diane Reeves – One for My Baby, Galactic – Black-Eyed Pea
  3. David Byrne’s American Utopia on Broadway
  4. The Band’s Visit on Broadway with Engineer: Kai Harada 
  5. Workbag: Earthworks, B&K, Studio6 Digital, Hilti laser measure, rubber mallet to bang on walls, a strategy for approaching projects
  6. Podcasts: Live From Here, Wait Wait Don’t Tell Me, TWIT “This Week In Technology”
  7. Books: Love is a Dog From Hell, Sound System Engineering, Sound Systems: Design & Optimization
  8. Quotes
    1. You don’t need to know why gasoline burns to drive a car, but it helps is you understand the fundamentals of how cars work and how they respond.
    2. Noise Criterion a series of curves where you make octave band measurements and what curve you stay under you use as your number.
    3. I think making rooms quiet makes them sound better. But if the show is 100dB then it doesn’t matter.
    4. It’s always cheaper to design it right in the first place.
    5. I’m a big believer in delaying the main system to the backline of rockbands.
    6. People are working to make the audio experience at concert venues like a movie experience.
    7. Because the transfer function inherently at mid and low frequencies is looking at the interaction of the room and the system and at high frequencies is looking at just the system I was hoping that as a tool, Smaart would bring those two things [acoustics and sound system design] together.
    8. The idea of low frequency decay being in some reasonable balance with high frequency decay in a room is critically important and a very important design tool and something that’s easy to measure in Smaart.
    9. My biggest pet peeve is people looking at the screen and not listening.
    10. If you have 80Hz as your crossover point, but your subwoofer is 6-8dB above the full range device, your acoustic crossover will be much lower than if you turn the subs up 10-12dB more. You’ve kept the electronic crossover, but slide the acoustic crossover up by changing the gain. I think you create a lot of mud in those cases, by having the subs go so much higher. I like to add EQ outside the bandpass on the subwoofer to make steeper crossovers and reduce the interaction in those areas.
    11. Complex FIR filters that address low frequencies introduce a lot of delay.
    12. I’m a big believer in delaying the main system to the back line of rock bands. So much sound is coming off of the stage that 7 or 8ms really makes a big difference for the front of the audience. The people up front stop hearing two snare drums.
    13. If you’re going to go out and optimize a system, you should have a step-by-step process in your head.

How to take fast impulse response measurements in Smaart© without pissing people off

By

Use these settings to efficiently investigate room modes with the impulse response module in Smaart©.

  • FFT: 128k
  • Avg: None
  • Delay: 0 (doesn’t matter)
  • Options > IR > Overlap: 0%
  • Signal Generator: Pink noise, Pseudorandom, Drop IR Data Window
    • Level: +10dB above noise floor

Measuring room modes sucks. It takes too long, pisses people off, and the ROI is debatable.

The last one is user dependent. Let’s work on the first two.

Why does it take too long?

This was user error on my part. I thought that multiple averages were required for sufficient dynamic range and actionable data. This meant that all of my measurements took twice as long because I had the averages set to 2.

While increasing averages does improve dynamic range, it is less important in researching standing waves because you’re going to average together at least 6 of them in the end.

Set averages to 0. Fixed.

Why does it piss people off?

  1. Wrong stimulus
  2. Unnecessarily high level

Use period-matched pseudo-random pink noise

The Pink Sweep stimulus pisses people off. People can get used to music and pink noise, but the sweep is surprising and painful. It’s the most common stimulus to enrage civilians.

While the Pink Sweep does improve dynamic range, the results I observed don’t justify the risk of someone telling you to stop. If you’ve got the room to yourself, though, do it.

From what I can see, using pseudo-random period-matched pink noise will get the job done at the same signal generator level and generate less complaints.

Set level +10dB above noise floor

I was using +20dB because according to the Smaart user manual you need to be at least 10dB above the noise floor, but 20dB is OK, and 30dB is preferred.

noise floor comparison

I ran through six tests at six different measurement positions:

  • A: Pink Sweep at +20dB relative to noise floor
  • B: Pink Sweep at +10dB
  • C: Pink Noise at +20dB
  • D: Pink Noise at +10dB
  • E: Pink Sweep at 0dB
  • F: Pink Noise at 0dB

Then I imported the batches of six measurements into Room EQ Wizard to be level matched at 100Hz, time aligned, and vector averaged. Below are the spectrogram graphs for each of the averages.

pink sweep vs noise at three levels
Spectrogram compare

Even though dynamic range drops as the stimulus in the room is lowered, I don’t see any difficulty in identifying the first four peaks in the first four measurements. Only when the stimulus gets to 0dB compared to the noise floor do I start to have doubts.

Obviously the results will be different in larger rooms compared to my living room, but going through all of these tests gives me the confidence to try a -10dB level moving forward.

Extra Credit

Level

Verify signal generator level by comparing it to the noise floor. Switch over to an RTA graph. Measure the ambient noise floor. Offset it +20dB. Turn on the signal generator and match it at every frequency.

While you’re there, make sure that the input levels (Mic/Ref) are matched and within the yellow target zone.

What about wind?

In any scenario where there might be a possibility of any significant time variance during the measurement period, you would probably be better off increasing the measurement time window and/or using a period-matched stimulus signal rather than upping the number of averages.

Rational Acoustics Smaart v8 User Guide

Signal Generator

Pink Noise – Pseudorando

Use a 2-channel measurement with period-matched signal generator.

At first you might wonder, why would I ever use a single channel measurement? IR measurements go back a long way and can be completed with a balloon and stopwatch. This direct method may deliver mixed results, though, since it might not have an instantaneous envelope or uniform spectral content and it doesn’t tell you anything about the sound system.

The newer indirect method uses a 2-channel transfer function to mathematically estimate the response and combined with a test signal of matched length can simplify our goal of actionable data.

feed the DFT what it really wants to eat: a test signal that either fits completely within the measurement time window or cycles with periodicity equal to the length of the DFT time constant. Signals that meet these criteria can produce deterministic, highly repeatable measurements in a fraction of the time it takes to get comparable results using random signals.

Rational Acoustics Smaart v8 User Guide

Unlike the Pink Sweep, pseudorandom cannot be triggered by the play button so you’ll need to start the signal generator before you press play or Smaart may crash, which is what happened to me the first time I tried it.

Pink Sweep

I was curious to see how the results of a pink sweep might differ so I tried that as well.

Using SATlive here to show IR overlay

Ld = level of direct sound

Ln = level of noise

Ld – Ln = dynamic range of the measurement

The dynamic range of the measurement using pseudorandom noise is -44.73dB. The dynamic range of the measurement using pink sweep is -65dB. That’s an improvement of 20.27dB.

FFT

Unless you are in a very large or noisy room, 128k (2730ms = 2.7s) or 240k (5000ms = 5s) should be plenty. I normally use 128k.

Averages

If we want better dynamic range of our measurements we can either turn up the signal generator in the room, increase the averages, or increase the measurement time. Since my goal is speed, I’m looking for the lowest number of averages and measurement time.

When measuring with period-matched noise or sweeps, averaging is normally set to “None” or 2, although it is still possible that a higher setting could prove helpful if measuring in an extremely noisy environment.

Rational Acoustics Smaart v8 User Guide
averages
AveragesDynamic rangeΔ
0-51.31
2-54.643.33
4-59.995.35

Overlap

Normally this is set to 0% for maximum noise reduction, but can be set higher to save time. Let’s see if there’s any benefit to using more overlap in terms of saving time or improving SNR.

This image compares an IR measurement with 0 averages and 0% overlap in yellow with 2 averages and 99% overlap in blue, which took about the same amount of time to measure.

Doesn’t look like much of an improvement.

Delay Locator

You only need to set the delay locator if you are using random stimulus.

The first is to delay the reference signal to match the timing of the measurement signal, so that the data windows line up. You should always do this when measuring with random signals.

Rational Acoustics Smaart v8 User Guide

We are going to use a period-matched stimulus so we can ignore this step. Yay!

Gratitude

Big thanks to Rational Acoustics support. My laptop crashed a lot during my first tests while “Processing frame”, which really sucked when the signal generator was stuck on. I emailed support. John emailed me back and I found out that an update with a fix was coming out that day. It worked!