Home Theater Forum and Systems banner

1 - 3 of 3 Posts

6 Posts
Discussion Starter #1
I have been using REW for quite awhile to try to tune my system EQ. At first I thought my mic (Dayton UMM-6) is broken, because it would measure all these 20dB+ nulls that the ears don't hear, but with time I learned that they're the inevitable result of room echoes comb filtering for full frequency range measurements. It's been said that sweeping one speaker at a time (rather than the stereo pair) takes care of comb filtering from minute path length differences to each speaker, but in my experience it didn't help much.

A few days ago I tried something different:
1. Generate a 30s white noise clip (padded with 5s of silence before and after) in Audacity;
2. Measure the system playing the clip with the mic--slowly wave the mic around in your listening area centering on your main listening position but also covering more peripheral places (such an approach would obviously be useless for a sweep measurement because the mic can only be at one place for any particular frequency)
3. (Optional) silencing out the measurements before and after the noise clip
4. Taking spectrum plots (Hanning, 65536 points) of the generated and recorded noise and exporting the results as separate text files
5. Opening the files in Excel and subtracting the recorded value of each frequency from the generated value (to remove the random variations of generated white noise)
6. Saving the result as .txt or .csv and importing as Frequency Response in REW

Now, this is obviously useless for anything but frequency response, and there's no algorithm rejection of the noise spectrum of the environment, but in my relatively quiet listening environment I got this result:
(Positional smoothed)
Whereas for the average stationary swept measurement I get something like this:

Now, the EQ and mic calibration for the two measurements are slightly different for the two measurements but I think you would all agree that this can be swept under the carpet compared to the differences in HF measurement consistency. Note that both measurements are unsmoothed.

Is there any way of adding a measurement methodology to REW that could work with a moving mic throughout the measurement? Change the source signal into an MLS sequence maybe?


Premium Member
1,562 Posts
Good find! I also use this type of process for EQ as well as others. Many experts question the value of this methodology, but it is still my favorite.

Here is a link to a thread discussing this subject >> MMM
1 - 3 of 3 Posts