Adventures with REW and DSP Correction

Hey All Fellow Forum Members,

I just got my hands on a copy or REW (Room EQ Wizard) analysis software and a mini-DSP UMIK1 USB microphone to evaluate my system in its current configuration and existing room treatment. This was prompted by John Darko’s recent YT video discussing RT60 (reverberation time with level attenuated by 60 dB). I was very curious to see how effective my “seat of the pants” room treatment scheme is working. Overall I’m very happy with the current SQ (sound quality), but you know how tough it can be to leave things alone and not “wonder” and attempt improvements.

Below is my very initial FR and RT60 response after a fair bit of fiddling with my LMS (loudspeaker management system) and sub tuning. This is really fun and informative, but the REW is also very powerful (and sometimes a bit confusing).
NOTE: My first attempts at FR capture showed massive SPL drop in the lower frequencies. This turned out to be a PC configuration for the mic that needed the “Enhanced Audio…” option unchecked. Then all was good.

You may notice this FR graph is both L+R and sub so I’m very curious to repeat with L and R separately. I think some of the “lumpiness” is due to comb filtering at the mic location (constructive and destructive interference) and I think that 9kHz dip must have something to do with this phenomenon. Will report back after L and R separate captures.

The RT60 curve looks good. General rule of thumb is that reverb time should not exceed around 500ms to 600ms from 200Hz to 2kHz.

Next up will be further exploration of the ETC (energy-time curve) which should be monotonically decreasing (so I’m told). Loads to learn and for just north of $100 for the mic this is one very useful tool.

7 Likes

I bought a Mini DSP mic a year or so back, but never got around to do room tuning.

4 Likes

That dip is really bizarre and quite severe. You have to figure out what about your room is causing that - do not try to correct it with DSP, it’s too huge. It could possibly be something to do with your listening position (assuming that was where the measurement was taken from) and/or speaker position. Have you put your room dimensions into a room mode calculator?

4 Likes

@badbeef I agree that 9kHz dip is huge and would not try to correct with DSP. I did a rough calculation of the wavelength at 9kHz and it’s around 1.5" so if you moved one speaker 0.75" that might correct for this. I also want to somewhat downplay what a single mic measures (L+R) versus my pair of ears separated by a few inches. If the L and R separate measurements look much better (much less attenuation) then I may simply chalk this up to “measurement error” or not truly representing my hearing experience.

Thanks for the input and would love to hear more from you and others especially those of you that have successfully set up studios and well-treated rooms.

1 Like

I am far from an expert, but this 9kHz strikes me as bizarre as well.

Separate left and right measurements may help determine what is going on.

2 Likes

If my memory is good, @brett66 & @vee have trodden this zone.

If it did, that would be surprising, but would tell you something. Another thing you can try is take a few measurements around where your head usually is, say a foot or so either way, L/R and front/back. If they vary wildly, it is the room and your seating position.

This Harman mode calc is pretty good as I recall - nice graphic representations rather than the usual vertical line charts:
https://www.harman.com/documents/Room%20Mode%20Calculator_0.xls

If your room isn’t too much unlike a regular rectangular box, this will tell you clearly where good and bad spots are, in all dimensions.

2 Likes

I have done this both with REW and Focus Fidelity and never seen that big of a dip at that high of a frequency. I agree with what someone suggested and try moving the mic a foot two in different directions. Then try moving the speakers. Correction will not fix that.

1 Like

This is true.

Unusual dip.

Measure each speaker near field (3-5ft away) to isolate.

4 Likes

It might help to have a photo of the room, as well as an understanding of the hardware and technique involved.

It’s an odd dip for both the magnitude and the frequency. I haven’t seen anything like it previously, but I could understand if that dip was affected by a mic not held in a proper stand, with furniture partially blocking things, and odd reflections coming off other furniture.

If you have leather furniture in the room - particularly near the mic - try covering it with a blanket.

What’s your technique? What’s the mic stand involved?

1 Like

I have used both REW and Acourate and settled on the latter as it is a single sweep from each channel and has a tool to help center the mic perfectly between the speakers at your listening position.

Both are very useful products. You need to be careful boosting any frequency more than 2db or so. The key is to bring all of the rest of the freqs down to match the lowest one within a few db. You have to be careful not to lower the corrected signal more than 10db or so. It’s a lot of fun and will help with SQ once you get a curve (or curves) that you like.

Those programs are also useful for technically integrating subs but I have not tried that function yet.

Be sure to take notes and give the files meaningful names or you will soon be lost in the ozone.

Here is my last uncorrected (smoothed) room sweep:

Without smoothing applied there will be some big dips and bumps but they are so narrow that there is no reason to try and correct them. Your large dip at the high freq tells me that something is wrong with the setup. Rooms wont usually affect the freqs in that high range due to the lengths of the wave.

1 Like

Thanks all for the advice and ideas. The mic is sitting just above the back of my leather couch on the tiny tripod included with the mic which is where my head usually sits while listening. Will try a blanket and L and R separate measurements near-field.

This is fun and really gets you thinking about ways to optimize room, etc.

This also feels very much like a bit of an art and not hard science since perfectly flat FR may not be the most enjoyable sound and as John Darko shared regarding RT60 measurement maximum times this is just a guide and varies with personal preference for how “live” you like your music playback.

If I remember right in REW there is a way to make several sweeps and include them in the same measurement. If you move the mic Left-Center-Right a few inches in front of your head, then L-C-R in the listening position at your ear location then L-C-R behind your head a few inches you can average these into one correctable signal…

Any way you do it you cant hurt anything and you will learn a lot about how sound waves move through the room.

We are really sitting inside the speaker (the room) and trying to make the experience as enjoyable as possible.

I use a corrected “house curve” that is 2db up at 20hz, 0db at 1khz, -6db at 20khz. Have fun and dont get lost in the weeds! :grin:

2 Likes

I totally agree with that.

I’m using a mini-DSP DDRC-88A, specifically for my subwoofers in Home Theater use, and I find that even after all the sweeps, adjustments and listening… different source materials have different sonic signatures. On my Harmony remote, I actually added two soft buttons to adjust the subwoofer (miniDSP) volume up and down, in case a particular show is boosting bass too much (and there are a lot of those!)

1 Like

That is exactly the process I used. The moving microphone method gives the best response for frequency response, it is similar to multiple microphone placement method, but far faster and IMO more accurate.

1 Like

Here’s an interesting vid right along this theme.

Brief update: Below is the much improved 15Hz to 500Hz FR range before any room correction auto-EQ from my LMS and after auto-EQ and quite a bit of optimization by manually changing the LMS and integrating the subwoofer by using the SVS app to add parametric EQ. The sub really helps flatten the lower frequencies and is quite effective at helping fill in the dips and attenuate a peak.

Also did some experimenting with higher frequencies to chase the weird 9kHz dip.

  • Moving the mic just an inch or two laterally caused the L+R FR to shift the high frequency dip by several kHz.
  • Measuring L and R separately shows no such dip at higher frequencies.
  • Adding a thick fleece blanket to cover my leather couch made little to no difference in the L+R high frequency nulls.

This suggests a strong interaction between L and R speaker outputs that are causing variable destructive interference at these higher frequencies. A.K.A. comb filtering as I understand it.

3 Likes

Early listening tests just now are hugely satisfying. Why did I wait this long to start getting my system more fully sorted? Without measurements this process would be totally aimless and random with suboptimal results for sure. The bass is fat, clean and foundational.

Next step is to add more bass traps for even better (lower) reverb times to clean up and even out FR for that bass further. I’m planning to use two GIK impression series 6" thick units with diffusion panels. They make a narrow version that will fit just above the windows on the front wall. And since they are at ceiling height this should make them more effective at “trapping bass”.

3 Likes

Just curious… how far behind mic is a wall? Do your measurements change when you stand directly behind the mic while taking the stereo measurement at MLP?

The rear wall is about 15 ft. behind the mic