Interesting discussion. When using pink noise I get the same results as I do when using sine sweeps regardless of speed of sweep, so not sure I understand what you're saying here?This is why I rely on pink noise and RTA to analyse a room instead of sine sweeps or pulses. As rooms have a fixed RT you can change the level of the pink noise and see the room response at the "saturated" levels you will be listening. I am sure a specific signal that is designed to work with a custom measurement system will do a better job.
Looking forward to the update
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Trinnov Altitude 16 Review (AV Processor)
- Thread starter amirm
- Start date
I didn't say they will be different. It is possible that depending on the room, the sweep speed and the level of the test signal, both methods may give the same results. But they may be different and the sine wave signal will be the misleading one.
No, it doesn't, though it's a common misconception. The requirement for capture of the test signal is to ensure that capture continues as a minimum for the 60 dB decay time after any part of the stimulus has been emitted. That isn't an arduous requirement since the sweep starts at low frequencies so the parts of the room's response that have the longest decays get the most time anyway, and capture continues after the sweep stops to ensure the requirement is met. It is more useful to think of the sweep as a stimulus and the capture as lasting long enough for the response to the stimulus to decay into the noise. The transfer function comes directly from the ratio of response to stimulus.
You have just said that it’s possible, if the test is not done correctly
I agree that if a room has an RT60 of 1sec a stimuli should be stationary for at least that amount of time. How long a 20,000Hz sweep should last then?
Last edited:
Trinnov has been hinting at improved bass management and modal correction for years now, always saying they would not change from their current scheme unless they could do substantially better, but that they were working on "something" in the back room. Be very interesting to see what they release.
On the measurement diversion: defining steady-state is of course a function of room size and listener position, but for most homes it is not an issue and the room correction folk (and others like REW) use sweeps proven to work in the real world. The times I have run into problems were in the past when I was helping acoustic designers for larger spaces, like our church, where the direct and reflected paths as well as reverb times were long enough to require longer dwell time in the frequency sweeps. The last time I did anything like that, I was using a fancy professional analysis package (whose name I have forgotten, sorry, it was not mine), and it had a neat mode that set the dwell time based upon room dimensions, RT60 (etc.) values it measured, and frequency. Longer dwell at 20 Hz, shorter at 20 kHz. I had (have, technically, but replaced by REW now) a "lower-end" analysis program costing only a few thousand dollars that starting each sweep with an "impulse" to help determine time delays and could also tell you if you needed to use a longer sweep for steady-state analysis (it did not do it automagically).
On the measurement diversion: defining steady-state is of course a function of room size and listener position, but for most homes it is not an issue and the room correction folk (and others like REW) use sweeps proven to work in the real world. The times I have run into problems were in the past when I was helping acoustic designers for larger spaces, like our church, where the direct and reflected paths as well as reverb times were long enough to require longer dwell time in the frequency sweeps. The last time I did anything like that, I was using a fancy professional analysis package (whose name I have forgotten, sorry, it was not mine), and it had a neat mode that set the dwell time based upon room dimensions, RT60 (etc.) values it measured, and frequency. Longer dwell at 20 Hz, shorter at 20 kHz. I had (have, technically, but replaced by REW now) a "lower-end" analysis program costing only a few thousand dollars that starting each sweep with an "impulse" to help determine time delays and could also tell you if you needed to use a longer sweep for steady-state analysis (it did not do it automagically).
The phrase "most rooms" is pretty woolly, don't you think? Besides, how were those REW tests made by layman proven to work? Naturally, when you have a peak of 20dB at 50Hz, whatever type of measurement you use, it will show on it. But we are on ASR where 1dB deviation of a speaker FR is the difference between poor and great. Why not use methods that will have much better resolution?
Of course, since there is no specific exact standard for consumer room size. But the biggest room in our house is maybe 20' in the longest dimension, pushing 30' including opening to the kitchen, and that is far less than the 150' or so of our church. IME, which may not at all match yours, most consumer listening rooms are smaller than that. If you want an acoustic definition of "small" versus "large" spaces there are plenty around based upon wavelength (frequency) and such. IIRC the difference I saw characterizing our church was in the 3 dB range or so depending upon dwell at certain frequencies, which was frankly more than I expected, and that was back-to-back measurements with nothing else changed (same SW, mic position, etc.) It turned out the initial default was very short and needing tweaking. As for the rest, argue with them, since they define the sweeps and I am just a bystander. I got some white papers many years ago for Audyssey and had a discussion with Dirac Live a few years later when they described their method; I had already had several long discussions with the acoustic engineer who developed the commercial SW package. I pulled a couple of Trinnov papers but have not really looked at them (actually one was a gov't study and another a university paper using Trinnov's HW). As for resolution, I did not mention that anywhere, and have no idea what method you are advocating. For me personally the gap between "poor" and "great" speakers is more than 1 dB, and more than frequency response, but I do not see how that plays into Trinnov's announcement.
I was not arguing with you or anyone else. I am simply saying that sinewave sweep measurement is not necessarily a good method.
I was advocating using pink noise and a RTA. I though I made that clear on my comment to @Frank Dernie that I made only today.
It doesn't, other than the subject is room measurement.
I'm not sure what you are agreeing with as that's not what I said. The sweep can be as short as you like as long as the capture isn't ended prematurely, a shorter sweep just means less energy in the stimulus and lower signal-to-noise ratio in the result. Large spaces are often measured directly with pistol or cannon shots, how long do you think they last?
First of all, gun shot is a pulse and the FR is calculated with FFT. It has nothing to do with frequency sweep measurements. Like the ones here. There is no capture windows on a sweep. The signal is a sinewave where the frequency is slowly increased.
Maybe you can elaborate the type of measurement you are talking about.
Ah, got it. Yah, it has its pros and cons.
I tend to pop in briefly whilst some sim or test is running so do not always get through the whole thread. I did not see that comment, sorry. Pink noise and an RTA is what I have usually used for room analysis, collecting samples over enough time to reach steady state and average the amplitude variations. Sweeps are nicely defined and controllable to set specific frequencies and such, though, unlike a noise source (though you can filter it and all that jazz). That said, doing a quick sweep and missing a large resonance is vexing, and doesn't usually happen with pink noise and longer monitoring.
All else aside, I want to visit the Trinnov site this weekend and see if any more detail is available. They have not offered any details or really even a broad notion of their plans until now.
As everything in engineering
It seems we are in agreement that RTA is a better option than sweeps. However, majority on ASR rely on sweeps.
The goal of the measurement is to determine the transfer function from source to mic. That transfer function can be expressed in various ways, most commonly an impulse response in the time domain or magnitude and phase responses in the frequency domain, both provide different views of the same information. If you google "Transfer function measurement with sweeps" (MÜLLER and MASSARANI) the first few pages have a review of the various methods through which a transfer function measurement can be made, including a variety of periodic and non-periodic excitations. Sweeps and impulses fall into the latter group. It reads to me as if you may have some misunderstandings of how sweep measurements are made and processed, the paper should help clear that up.
It is stunning how we fail to communicate. You are agreeing with me but still asking me to learn what I am advocating. Maybe you need to pay attention to the communication?
Meanwhile, thank you for your confidence in me and my education
For me, and perhaps only me, it depends upon what I am doing. Sweeps, perhaps gated, are a nice way to check out the frequency response of a system, and work well for component measurements (vs. speaker response). Pink (or whatever color) noise is nice for in-room acoustic measurements and getting a broadband level response, but does not provide time (delay) or reverb info. Both require some fiddling with the analysis tools, natch, like FFT lengths and so forth if you are using them for SPL or RTA measurements.
The "professional" tools I had did impulse, colored or white noise, and sweeps under user control. In the past pink noise and an audio spectrum analyzer served me well and meant I did not need a tracking signal generator. Most folk using SW like Multitone or REW do sweeps, I think, though I think REW has a noise source as well (have not looked in a while, just did sweeps most of the time). Most of the time these days I do a sweep in REW, because it is easy and what everybody else does most of the time. Lazy... But a proper sweep provides a lot of useful information and control you don't get with pink noise.
For room analysis, the programs I have used seemed to mix pink noise and sweeps during their analysis. I usually spend a week or two fiddling when I get a new component and then don't touch it again unless something changes so I've lost track of which program does what.
None of this is my day job, though I maintain interest, and try to keep some semblance of competence.
Last edited:
I'm not. You are advocating measuring a room's response with pink noise and an RTA. Even assuming you mean periodic pink noise rather than random, it is a method that provides a very limited view of the room's behaviour, with no temporal information at all. A sweep can provide a complete view of the room's behaviour in less time and with higher resolution. Comments such as "There is no capture window on a sweep" are indicative of a lack of understanding of sweep measurements. The paper may help, if you care to read it.
As Insaid earlier, thank you for caring to understand who you are talking to. You have been very rude. I do not want to continue chatting with you.
OP
- Thread Starter
- #1,079
I want to caution you to be cordial when communicating with our technical experts. They have earned their tag for a reason. Disagree all you want but don't get personal.
- Joined
- Jan 1, 2020
- Messages
- 1,384
- Likes
- 2,107
Unfortunately for him, he’s a repeat offender and he’s also the only individual on my ignore list here since he does not care to have a civil conversation.
Similar threads
