-
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!
UMIK-1, REW adventure
- Thread starter Ron Texas
- Start date
From the help:
REW allows multiple sweeps to be averaged, although best results are generally obtained by using single, longer sweeps rather than multiple, shorter sweeps. Do not use multiple sweeps if the input and output are on different devices (for example, if the input is a USB mic). If Sweeps is more than 1 REW uses synchronous pre-averaging, capturing the selected number of sweeps per measurement and averaging the results to reduce the effects of noise and interference. The pre-averaging can improve S/N by almost 3 dB for each doubling of the number of sweeps. Averaging can be useful if the measurements are contaminated by interference tones, whether electrical or acoustic, as they typically will not add coherently in the averaging and hence will be suppressed by the process
Warning: some soundcards do not maintain sample synchronisation between the successive sweeps which produces a corrupt measurement that has multiple, closely-spaced peaks of approximately the same level in its impulse response, 1 peak for each sweep. This can also happen if the input and output are on separate devices. If the frequency response with multiple sweeps is significantly different from the response with a single sweep, stick with single sweeps
REW allows multiple sweeps to be averaged, although best results are generally obtained by using single, longer sweeps rather than multiple, shorter sweeps. Do not use multiple sweeps if the input and output are on different devices (for example, if the input is a USB mic). If Sweeps is more than 1 REW uses synchronous pre-averaging, capturing the selected number of sweeps per measurement and averaging the results to reduce the effects of noise and interference. The pre-averaging can improve S/N by almost 3 dB for each doubling of the number of sweeps. Averaging can be useful if the measurements are contaminated by interference tones, whether electrical or acoustic, as they typically will not add coherently in the averaging and hence will be suppressed by the process
Warning: some soundcards do not maintain sample synchronisation between the successive sweeps which produces a corrupt measurement that has multiple, closely-spaced peaks of approximately the same level in its impulse response, 1 peak for each sweep. This can also happen if the input and output are on separate devices. If the frequency response with multiple sweeps is significantly different from the response with a single sweep, stick with single sweeps
Do you use mixed-phase compensation (--mixed) with PORC?
It's been a while since I generated the filters, but I think to remember that I did use mixed-phase.
thanks for the help, now the filters sound fine. I have still to do much more work, but at least I'm not doing as many gross mistakes as before.
How do you apply EQ en REW: using the invididual measurements per channel, or using the Vector Average curve of the left&right channels?.
Another silly question, I'm still unable to measure the response with the filters. I want to have real measurements of the result with the microphone, not calculated curves. The generator tab does does not read any corrected response, and later one would have to measure that impulse corrected.
How do you apply EQ en REW: using the invididual measurements per channel, or using the Vector Average curve of the left&right channels?.
Another silly question, I'm still unable to measure the response with the filters. I want to have real measurements of the result with the microphone, not calculated curves. The generator tab does does not read any corrected response, and later one would have to measure that impulse corrected.
- Joined
- Apr 8, 2019
- Messages
- 163
- Likes
- 211
Ron,
Point the mic between the speakers and use the zero degree calibration file. Ceiling pointing is more for surround sound measuring.
Here are some areas to fine tune different things in REW. I think you did it right btw.
First - try a cut only filter - in the 4th or 5th EQ tab set the max gains all to 0 db. This will only make cuts instead of boosts to frequencies.
Second look at your curve pre correction and match the target curve closer to that instead of the default. In my room I have high treble roll off because everything is firing into big plushie theater chairs...REW will try to boost the crap out of this to meet a flat default target curve. Instead I just make a curve that follows it down and it sounds amazing...+8 db boosting from 3k to 20k doesnt not sound amazing lol.
If you have room, try to get the response as flat as possible with positioning, then measure and correct. I like to eliminate SBIR as best I can or at least put it under the crossover frequency of the mains then wall load the subs to deal with it.
Definetly take more than one measurement and make sure the mic is excactly centered between the speakers - ie 100 inches to each tweeter from the mic position or whatvever. (on a boom mic)
Point the mic between the speakers and use the zero degree calibration file. Ceiling pointing is more for surround sound measuring.
Here are some areas to fine tune different things in REW. I think you did it right btw.
First - try a cut only filter - in the 4th or 5th EQ tab set the max gains all to 0 db. This will only make cuts instead of boosts to frequencies.
Second look at your curve pre correction and match the target curve closer to that instead of the default. In my room I have high treble roll off because everything is firing into big plushie theater chairs...REW will try to boost the crap out of this to meet a flat default target curve. Instead I just make a curve that follows it down and it sounds amazing...+8 db boosting from 3k to 20k doesnt not sound amazing lol.
If you have room, try to get the response as flat as possible with positioning, then measure and correct. I like to eliminate SBIR as best I can or at least put it under the crossover frequency of the mains then wall load the subs to deal with it.
Definetly take more than one measurement and make sure the mic is excactly centered between the speakers - ie 100 inches to each tweeter from the mic position or whatvever. (on a boom mic)
OP
- Joined
- Jun 10, 2018
- Messages
- 6,223
- Likes
- 9,346
- Thread Starter
- #28
I have been doing the cut only adjustments for a while and have been experimenting with different target levels to warm things up a bit. This time I checked for exact centering and used the horizontal calibration. Results are a bit different.
I generated filters from my last impulse response measurement yesterday using the --mixed setting and the B&K target. Sounds pretty good, perhaps more natural than my DRC-FIR filters. I haven't had a chance to measure the result, though. Have you tried DRC-FIR?
No, at the time I tested various systems I hadn't known about DRC-FIR. I tried REW, Dirac Live, Sonarworks 3, IK Mulitmedia Arc, and PORC. With PORC (using REW for the measurements) and Dirac I heard the biggest improvement. For me it's mainly the soundstage that snaps into focus with these PORC filters. The phantom middle and soundstage are more defined. The instruments positions are a bit blurry without them, like a watercolor painting. With the filters the instruments are more clearly separated, and don't blur into each other. Hard to describe, but it's readily audible.
In any case I found it well worth to invest some time and investigate the options. Depending on your room and equipment it can bring a clear improvement.
One thing I haven't figured out is how to deal with higher sample rates. I get a scalloping effect on the higher frequencies. My impulse response is measured at 48000 Hz and I resample with sox for the other rates. Downsampling on the fly would be fine with me if the final result sounds good, but I haven't figured out how to do this with brutefir and LMS yet.
Test convolution at 44100 Hz:
96000 Hz:
Also, you have to specify the target explicitly for the higher sampling rates.
Strange. I used Sox too, but I don't remember having any problems with sample rates. What didn't work for me at first was the target curve. No matter which I selected, the correction was always flat. I think it was a problem with the code, and the bstegmaier repo I linked to fixed that. I think in the end I went with B&K2.
Don't know if it is of any help, but try to make a few more measurements with REW, and then generate filters with them. For some reason the results were always a bit different for me, even if the curves looked pretty similar. Some worked better than others.
Don't know if it is of any help, but try to make a few more measurements with REW, and then generate filters with them. For some reason the results were always a bit different for me, even if the curves looked pretty similar. Some worked better than others.
Ah, I was using a different repo, greenm01. I'll try the other one.
I just did the following measurements with REW: (1) Left channel, (2) Right channel, (3) Left+Right channel. The result of doing Vector Average on (1)&(2) was very close to (3) (excepting some expected vertical shift for the SPL). Very nice.
I keep struggling on how to measure the filters created by REW. I've seen that people use a miniDSP for that, but looks overkill.
Now, with the impulse filter response on foobar the music is less boomy, alghough I don't know if I like it more.
BTW the biggest benefit I've found with REW+umik+laser meter is to place much better my monitors. Now I get similar SPL from left/right speakers in the middle/trebble. Before was not the case.
I keep struggling on how to measure the filters created by REW. I've seen that people use a miniDSP for that, but looks overkill.
Now, with the impulse filter response on foobar the music is less boomy, alghough I don't know if I like it more.
BTW the biggest benefit I've found with REW+umik+laser meter is to place much better my monitors. Now I get similar SPL from left/right speakers in the middle/trebble. Before was not the case.
What are the bennefits using filters. With mathaudio room eq you don't need creating/using filters ( i don't know mabey Mathaudio creat them automaticly on the background) it measure everything between 20 an 20.000 hz an let you self adjust the vertical line by a slide bar. Even so you can adjust however you see fit this horzontal line. I'm quite curious if you will hear any difference between rew an mathaudio. The reason i use Mathaudio is because it is way more simpel than REW. An i can adjust my slope however i see fit somthing you probably can't do with REW ( not sure). Have look at the measuring points in mathaudio https://mathaudio.com/images/measurement-points.png
This is flat choice.
This is my personal choice which sounds more dynamic. Sometimes i change back to the flat choice esspecialy with jazz like Chet baker or Ahmad Jamal an others.
Last edited:
Benefits?. More likely, only wasting me time. I'm just using REW to learn and later use more sophisticated DSP, as rePhase. At this moment, I just want to know what expect from room correction in my setup, and that will probably guide me in future spending.
I'm less than a novice to understand REW (filters how to setup my mic spl etc etc) fully. So when i came across to Mathaudio Room EQ i found the simplicity a relief. I was up an runnig in minutes and the difference was massive an i could adjust the slope to my personal choice don't need specific hardware and or costly software an it is free for Foobar2000. I'm realy wondering if the difference between REW an Mathaudio is worthwill if so i will gather the courage how to master REW.
Last edited:
Cloned https://github.com/bstegmaier75/porc , but I'm seeing the issue with 96kHz. One thing I never mentioned is that I had to make several fixes to the code to get it to run with a recent numpy and scipy (most changes related to floats being used for indexes, which is sloppy coding IMO). So maybe my changes caused the problem. In fact, I tried running the original code with an Anaconda distro going as far back as their archive goes, but it still was not old enough. I'll have to try tracking down an numpy/scipy/matplotlib old enough.
It's a pity that it isn't developed further anymore. Can be tricky - or impossible - to fix problems in the code. I hope you get it sorted out.
porc is full of bugs, don't waste your time. Use better Rephase.
Results from my hifi experiments:
1. Filter from REW removes a room mode at 80 hz (as per calculated response, I've been unable to measure the effect of the filter), and the sound
gets OK, but not something to enjoy.
2. Using measure and filter from REW to create a phase corrected impulse in Rephase is definitevely much better. Phase is important.
3. Sound is more focused with the phase corrected impulse, but no sure if I prefer it to plain uncorrected sound. It's kind of focused vs. boomy and natural.
4. My best results so far are using rew+rephase filter and adding on top of that a low bost EQ.
5. Definitively, I like bass ;-). So my only actionable result: my next monitors must have a larger bass driver than my current Proac 1SC.
ah!, I have to say this: I'm listenning to a jazz radio channel in Tunein on my crappy 200$ cell phone at night with a good beer, and I am enjoying it more that my DAC+amplifier+monitors in my office along the day. So, for me: type of music and state of mind is more important for enjoyment than gear.
Results from my hifi experiments:
1. Filter from REW removes a room mode at 80 hz (as per calculated response, I've been unable to measure the effect of the filter), and the sound
gets OK, but not something to enjoy.
2. Using measure and filter from REW to create a phase corrected impulse in Rephase is definitevely much better. Phase is important.
3. Sound is more focused with the phase corrected impulse, but no sure if I prefer it to plain uncorrected sound. It's kind of focused vs. boomy and natural.
4. My best results so far are using rew+rephase filter and adding on top of that a low bost EQ.
5. Definitively, I like bass ;-). So my only actionable result: my next monitors must have a larger bass driver than my current Proac 1SC.
ah!, I have to say this: I'm listenning to a jazz radio channel in Tunein on my crappy 200$ cell phone at night with a good beer, and I am enjoying it more that my DAC+amplifier+monitors in my office along the day. So, for me: type of music and state of mind is more important for enjoyment than gear.
Last edited:
Similar threads
