Soundman2020 wrote:Source of the post Dr Space wrote:Source of the post Here are the results. I hope Stuart is around as can give his feedback...
Joules says the REW free ware is flawed
Well, sorry to disagree, but REW is not flawed: it is an excellent tool, very powerful, very well written, better than many paid applications, very well regarded among acousticians around the world, and is very accurate. What specific "flaws" is he referring to? Making general accusations without specifics is sort of strange and not very useful. If he is aware of such flaws, he should probably talk directly to John Mulcahay, and ask him to fix them. John is very responsive to comments and bug reports about the software, and would likely fix them quickly, if indeed there are any flaws.
Hello Stuart.. First, thanks for your feedback... I did not want to get into it with Joules. He has his own strong opinions about a lot of things. I can ask him more details about what he thinks is wrong..
so he uses some other package to make the measurements.
Which one? I don't recognize it. The waterfall plot looks vaguely like the old ETF system, but I think development on that stopped several years ago, and some of the graphs look very different in any case. It would be good to know which app he uses, and which specific flaws it does
not have, that he thinks REW
does have.
I will ask him what the software is.
Anyway, after some calibration stuff, he made 3 measurements with a volume around 80-90db
If that's the case (which is, indeed, the "standard" level for doing such measurements), then why do the graphs show levels about 10,000 times higher, at around 120 dB? In fact, both the Frequency Response and waterfall graphs show some peaks approaching 130 dB! Did he not calibrate the system, to show the real SPL levels? It isn't necessary to calibrate if you just want to compare relative levels, true, but for looking at absolute levels it
is necessary...
he made 3 measurements front of the room. mix position and rear of the room.
Which of those three is shown in the graphs you posted? I'm guessing they are for the mix position? Also, are those graphs all from the left speaker, all from the right speaker all from both speakers, or from some other combination?
We only have the left speaker connected at the moment. I also assume they are also from the mix position.
Here are the results.
My first impression is that the smoothing on the graphs is rather extreme, and possibly hiding important details. using 1 octave band smoothing on the RT (decay) graphs tells you very little at all, and the international specs commonly used for control rooms (such as EBU-Tech.3276 and ITU-BS.1116-3) call for measurements to be done at one-third octave resolution, with no more than +/- 50ms variation between adjacent bands. I'd suggest that you ask him to show that graph at the "standard" resolution, so you can see how the room is really performing, as compared to the specs. Here's what a typical RT graph should look like. This is for a room a little smaller than yours, shown at the normal spec resolution (1/3 octave smoothing): SNOLO--REW-RT60--50hz-12khz-FINAL--1..3+cursor200ms.png
Next impression: The room is rather dead, overall, with very short decay times for the size and purpose. The graph shows decay times of around 140 ms. For that size room, the specs call for something more like 210 ms (according to ITU/EBU specs) or 240 ms (according to IEC/AEC specs). Those numbers are overall RT times: specifically for the mid range, the specs are 260 ms between 200Hz and 4kHz for ITU/EBU, and 290 ms according to IEC/AEC specs. So your room is about half of what it should be, according to commonly used international specs for control rooms and similar critical listening rooms. So I'm assuming that there is still some acoustic treatment missing, to return the missing energy to the room? Otherwise it will be unpleasantly dry, and tiring to work in for long periods.
On the waterfall graph, there seems to be some type of strong modal resonance at a little under 120 Hz, which is unusually high for a treated room, so once again I'm assuming that some treatment is missing from the side walls (since that would probably be the 0.0.3 mode). The scale of the graph doesn't allow for closer scrutiny, but it seems that the decay time there is considerably longer than anything else in the room. Here too, the scale is overly compressed/smoothed to be useful. Here's what a typical waterfall plot looks like for a room similar to yours, at two different smoothing levels (first 1/6 octave, then 1/48 octave):
SNOLO--REW-WF--20-230-FINAL--1..6.png
SNOLO--REW-WF--20-230-FINAL--1..48.png
You can see that smoothing too much can hide important details. This is in a well treated room, but there are still some details visible on the higher resolution graph that could do with addressing.
The impulse response graph is not much use in the form shown in the graph you posted: the scale is way to compressed to discern any detail, but it does seem that you have some strong early reflections at around 15-18 ms, with lesser reflections out to about 40 ms. Here again, I'm assuming that additional treatment is planned but not yet installed? This is what a useful IR graph typically looks like, at the correct scale: SNOLO-REW-FIR--190ms.png
As Glenn mentioned, the frequency response graph looks reasonable, but it has been smoothed to 1/6 octave, so no real detail is visible in the lows and low-mids. I normally use 1/24 octave for those regions, which are critical for overall room response. Yes, it is true that the specs only call for 1/3 octave analysis, but that is never sufficient for a high quality control room. It is very easy to miss major issues by only looking at 1/3 octave smoothing. Here's an example of a well-tuned small room, similar in size to yours, at two different resolutions. 1/3 octave, and 1/12 octave. You can clearly see the difference around the 1 kHz mark. The smoothing hides that, makes it appear less serious.
FrKCAUS-REW-FR-12-22k--1..3-FINAL.png
FRKCA-US-REW-FR-12..22k--1..12--FINAL.png
>So, overall, there still seem to be issues with the room that need addressing, and as I mentioned several times, I'm assuming that there is >still treatment to be installed to deal with those.
Yes and no.. All the walls are sealed and the fabric frames are on so it might not be that simple but we will do what we have to do, I hope but it will probably not be Joules as he has other jobs he is under pressure but I will do the measurements with REW and try to do what I can with friends.
In summary: the scales on the graphs are not very useful to actually see what is really happening with the room. Good for a rough first approximation maybe, but not for precision tuning for a high-end control room. One other thing: there is no phase graph, so it's hard to say what's really happening in that domain. That is also very useful information, for room tuning. If you want, get him to export a true impulse response data file, send it to me (or post it here on your thread), and I'll do a more detailed analysis.
Here is what I wrote to Joules. This should allow us to learn a lot more if he provides this..
"He would like you to export the true impulse response data file so he can analyze this and also to know the name of the software you are using. Thanks Joules...
Could you also create a new frequency response graph smoothed to 1/24 instead of 1/6 octave as this will reveal much more detail. The same for the RT graph he said the industry standard is 1/3 octave resolution and smoothing at 1 octave is hiding a lot of details so if you do this at standard resolution this would be very helpful to see how the room is really performing.."
Thanks folks for all the feedback..
scott