Perhaps you could become better acquainted with "smoothing" and "windowing" before diss'ing others. Or maybe I should.
B.
I hope I didnt come across as too harsh. sorry about that. sometimes I get defensive when called a liar
the mmm measurements dont require smoothing. thats the beauty of them. They can be used as is because they are not subject to the dropouts of a single position measurement.
MMMs are also amazingly highly repeatable. after 15+ seconds you just are not going to see deviation in a forever averaging measurement unless you add noise or something external.
if you look at the Anechoic data(black line), you will see its Jagged. any smoothing or windowing would have removed that.
so I will reiterate, the measurements I posted do not have any smoothing or windowing applied. zero zilch nada
I can send you the mdat if you want.
Last edited:
I think I understand, I just disagree
I do know that making my system as colorless and "accurate" as possible is the best chance that I have to obtain the producers intent.
Sure there are lots of problems with recordings, but "good" producers understand those and work with them. It would be ill-advised for me to work against them.
I agree with one caveat. It is the best chance to obtain, on average, the intent of various producers. Measurements of mixing and mastering suites show wide deviation, and then there's taste. But on average, a neutral system will get you the best results across a large number of recordings.
Very interesting topic.
In fact I happen to agree with much of the white paper just in my own experience.
I think folks are gathered in two camps , one that want to quickly get there loudspeaker perform with minimal effort or steps to take , and the other folks that like seeing screenshots and analysis and like looking at graphs more than listening to the speaker.
I measure a lot in cars and the moving mic with plain old pink noise can get things spectral fast and the times I've done sweeps and averaged told me more about the room and any adjustment I made wasn't as refined and smooth sounding as a moving mic pink noise method. I think it's further suggests why a lot of folks have gone to auto eq in REW because it accepts both types of averages. And it's fast n easy , but auto eq still doesn't sound quite as good as a moving mic RTA.
Now in some cases we need to analyze the IR. Filter building for starters and room mode correction as well, but speaking strictly from a EQ standpoint and fast results moving mic RTA really seems the most repeatable way to achieve good results.
cheers
In fact I happen to agree with much of the white paper just in my own experience.
I think folks are gathered in two camps , one that want to quickly get there loudspeaker perform with minimal effort or steps to take , and the other folks that like seeing screenshots and analysis and like looking at graphs more than listening to the speaker.
I measure a lot in cars and the moving mic with plain old pink noise can get things spectral fast and the times I've done sweeps and averaged told me more about the room and any adjustment I made wasn't as refined and smooth sounding as a moving mic pink noise method. I think it's further suggests why a lot of folks have gone to auto eq in REW because it accepts both types of averages. And it's fast n easy , but auto eq still doesn't sound quite as good as a moving mic RTA.
Now in some cases we need to analyze the IR. Filter building for starters and room mode correction as well, but speaking strictly from a EQ standpoint and fast results moving mic RTA really seems the most repeatable way to achieve good results.
cheers
I'm interested in this. The mic movement is supposed to be slow, but how slow would it need to be to work with the sliding window? Or is there some other method that might be better than a simple average?
Based on my limited experience I can say the reproducibility of the MMM technique is remarkably high. Assertions that waving a microphone around can't yield reproducible results aren't supported by my experience. I'd be interested to hear from those who have actually tried it but have drawn different conclusions.
As for the question of whether the MMM results provide a good basis for equalization, I've tried equalizing the early arrival magnitude response, and equalizing the MMM (magnitude) results. After listening to music I've been happier with the MMM results so far, but I need to do a more sophisticated job with the first arrival approach. I've tried averaging impulse results over multiple measurement positions, which seems to help, but I don't think my averaging approach would hold up to critical scrutiny.
I'd like to try equalizing for maximally clean CSD response, over a reasonable listening volume, but haven't yet figured out a convenient way to do that. I'm still a pretty new user of REW so I may be misusing it or missing some of its features.
In any case, I find the MMM approach, at a minimum, to be a very useful reality check on my measurements. It provides a quick way to avoid going down rabbit holes trying to equalize very local effects.
Few
As for the question of whether the MMM results provide a good basis for equalization, I've tried equalizing the early arrival magnitude response, and equalizing the MMM (magnitude) results. After listening to music I've been happier with the MMM results so far, but I need to do a more sophisticated job with the first arrival approach. I've tried averaging impulse results over multiple measurement positions, which seems to help, but I don't think my averaging approach would hold up to critical scrutiny.
I'd like to try equalizing for maximally clean CSD response, over a reasonable listening volume, but haven't yet figured out a convenient way to do that. I'm still a pretty new user of REW so I may be misusing it or missing some of its features.
In any case, I find the MMM approach, at a minimum, to be a very useful reality check on my measurements. It provides a quick way to avoid going down rabbit holes trying to equalize very local effects.
Few
Not a different conclusion, but a caveat. I was getting very unrealistic results when I was sweeping too fast. I think that for repeat-ability that a maximum sweep speed needs to be established.
The other factor I would say is that you need to sweep long enough for the average to stabilize. I found when I did a short sweep that putting down the mic and reaching for the stop button caused the results to vary considerably.
Provided you always sweep slowly enough and for long enough, then the results should be repeatable.
Tony.
Tony--I've found the same thing. You have to be careful not to end the measurement prematurely. i've not done much study of the sweep speed but I can imagine that being over-eager could yield spurious results.
Pano: I agree that mult-point averaging should yield similar (but perhaps not identical) results but I find it much harder to reproduce the averaged results than the MMM results. That doesn't make them less relevant, but it does make them more tedious to make.
Few
Pano: I agree that mult-point averaging should yield similar (but perhaps not identical) results but I find it much harder to reproduce the averaged results than the MMM results. That doesn't make them less relevant, but it does make them more tedious to make.
Few
If you had done that test yourself you would know for a fact that the curve would be the same except one curve is 2dB higher.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.