First attempt at DSP/Room Correction with MiniDSP SHD Studio, REW, and DIRAC Live

[ctakim]

So had a chance this weekend to start to do a little DSP and Room Correction on my office near field desktop 2.1 sound system. First, I will list the components. I have recently purchased Genelec 8341a speakers set up as an equilateral triangle on my desktop with 4 inch hard foam blocks as speaker stands, they are being digitally fed via AES/EBU inputs coming from an also new MiniDSP SHD Studio. The source material is USB output from my MacBook Pro or my MacMini or Roon/Qobuz over the Gigabit ethernet LAN. The SHD Studio outputs 1+2 are feeding the Genelecs, and Output 4 goes to the sub via SPIDF digital output to a Schiit Modi 3+ DAC with analog connections to a single REL T7i sub (no hate please). A picture of my desktop setup is here:

Your Desktop Audio Setup?

MacBook Pro/MacMini-->MiniDSP SHD Studio --> Genelec 8341A

audiosciencereview.com

The REW frequency response readings from this system as is with no crossover (full signal to all speakers including the sub) is shown below. This is an average of seven microphone positions bracketing my listening position when I sit in my desk chair with 1/24 smoothing. I'm showing this as the starting point for my DSP room correction efforts.

Not surprisingly, the Genelecs have a pretty smooth frequency response at the higher end but a fair amount of unevenness in the subwoofer range, with substantial role off starting at 40 Hz and below. Parenthetically, there really is nothing resembling room treatments as this is my home office space. Just a bunch of boxes and file cabinets nearby.

 

[abdo123]

How far are the speakers are from the back wall?

I would limit the correction to 250Hz in Dirac and call it a day.

 

[ctakim]

So I started with the REW to check a few other positions for the sub, but ended up putting it in the left corner based on the frequency response curves. Oh, and I should mention that I'm using REW 5.20.5 and a MiniDSP UMIK-1 to do my measurements. I explored a few different crossover points and time delay settings and after some basic trial and error settled on 100 Hz as the crossover with a relatively steep falloff using the BW 24dB/oct filter for both the sub and the mains. I also gave the sub a 2.9 ms delay. This seemed to give a relatively consistent FR at the crossover point. With those settings programmed into the MiniDSP SHD Studio, I then fired up the DIRAC Live 3.2.3 software and let it do its thing using the standard House Curve. It was very easy to use and with all the settings in place and the DIRAC Live processing in place, the new FR curve now looks like this:

Not perfect as there still is some roughness 100 to 400 Hz range but also not bad for a first attempt at DSP and room correction! Now to spend some time doing some subjective listening!! I also plan to explore some different DIRAC Live House curves including those with a bigger bass boost (thanks to Youthman!!).

 

[ctakim]

How far are the speakers are from the back wall?

I would limit the correction to 250Hz in Dirac and call it a day.

Only about 1 ft (0.3 meters), which is not ideal but there is no way I can move the desk out into the middle of the room.

 

[Webninja]

Actually, per Genelec, that is an ideal distance. Although I don’t have Genelecs, I’ve been using their speaker placement guide for my desktop actives.

Monitor Placement - Genelec.com

To create the ideal listening experience, several aspects need to be taken into account when placing the monitors within the listening space. First, you need to identify the listening area within the space and consider where the listener will be placed in relation to the space as well as the...

www.genelec.com

 

[abdo123]

.

Only about 1 ft (0.3 meters), which is not ideal but there is no way I can move the desk out into the middle of the room.

Then adjust the target curve in Dirac so the null between 160Hz and 250Hz is not boosted.

 

[ctakim]

.

Then adjust the target curve in Dirac so the null between 160Hz and 250Hz is not boosted.

I'm not sure I understand what you are suggesting. Why is a flat response not to be desired over having a null dip in that region?

 

[RayDunzl]

I'm not sure I understand what you are suggesting. Why is a flat response not to be desired over having a null dip in that region?

If the null is due to cancellation - standing wave is out-of-phase with itself, adding energy won't fix that listening spot - it will still null - but may cause too much energy at that frequency in other areas of the room.

I have an un-fillable hole when both speakers play the same frequency with the same phase, around 48Hz. If the signal is stereo (not the same phase) around that frequency, there's no hole, because the standing wave cancellation has moved to some other location in the room,

I didn't even know it was there, until measureing. Somehow, it isn't obvious in music.

Also, there is a narrow dip at 220Hz, the speakers are dipoles, and cancel that frequency from the bounce off the wall behind the speakers.

Changing the phase, by delaying one speaker (experiment), changes the response in that area:

Which proves (to me) it is a cancellation null, and can't be filled with more amplitude.

---

Using the RTA window, observe the measurement for different pieces of music, and see when the dip does and doesn't occur, due to differences in the signal in that area.

 

[Hipper]

So I started with the REW to check a few other positions for the sub, but ended up putting it in the left corner based on the frequency response curves. Oh, and I should mention that I'm using REW 5.20.5 and a MiniDSP UMIK-1 to do my measurements. I explored a few different crossover points and time delay settings and after some basic trial and error settled on 100 Hz as the crossover with a relatively steep falloff using the BW 24dB/oct filter for both the sub and the mains. I also gave the sub a 2.9 ms delay. This seemed to give a relatively consistent FR at the crossover point. With those settings programmed into the MiniDSP SHD Studio, I then fired up the DIRAC Live 3.2.3 software and let it do its thing using the standard House Curve. It was very easy to use and with all the settings in place and the DIRAC Live processing in place, the new FR curve now looks like this:
View attachment 198986

Not perfect as there still is some roughness 100 to 400 Hz range but also not bad for a first attempt at DSP and room correction! Now to spend some time doing some subjective listening!! I also plan to explore some different DIRAC Live House curves including those with a bigger bass boost (thanks to Youthman!!).

That certainly looks pretty good - still 1:24 smoothing I assume.

As the key area looks to be the 0-400Hz region, can you post a graph of that with 'no smoothing'?

The main thing though is that you give this new arrangement a good listen over at least a week or two to get used to it. You may find initially that the bass sounds weaker then you are used too but eventually you might realise that in fact it is much more musical and allows the higher frequencies to shine through too.

 

[ctakim]

That certainly looks pretty good - still 1:24 smoothing I assume.

As the key area looks to be the 0-400Hz region, can you post a graph of that with 'no smoothing'?

The main thing though is that you give this new arrangement a good listen over at least a week or two to get used to it. You may find initially that the bass sounds weaker then you are used too but eventually you might realise that in fact it is much more musical and allows the higher frequencies to shine through too.

Thanks, that measurement curve above did have 1/24 smoothing, but here it is with no smoothing. Note that this is the average of 7 measurements that bracketed the relatively small area of where my head is when I sit at the desk in front of the computer screen and speakers.

Also I'm playing around with different House Curves with with different levels of bass boost from 4, 8, to 10 dB. I got these from Youthman's website here:

https://youthmanreviews.com/harman-target-curves-for-dirac-live/

But straightaway the 8 and 10 dB Bass boost sounds way to boomy to me. The 4 dB sounds a bit more acceptable while the flat DIRAC Live standard curve sounds a bit weak in the bass region, which is what you predicted. But you are right that I should give it an extended listen before making a long term decision. The nice thing about the MiniDSP SHD Studio is that is can store four different DIRAC Live presets that I can change or switch between with a press of the button!

 

[Hipper]

That's an exceptionally good looking curve for no smoothing with dips of only 2 or 3db off the average slope by the look of it. I doubt you can do much better then that.

Presets are good although I don't use them in my situation. I have some CDs that have more bass then others for example.

The general aim seems to be to get a smooth frequency response (FR), which you've got.

One thing I found as equally important was decay times. You can check how you are doing with this with REW using Waterfall plots or a spectogram. These can be found on the tabs once you've opened a measurement. You are looking again for smoothness. My decay times are around 200ms through most of the frequency range. Some say this may be a bit 'dry'. I would think between 200-300ms would be good. Most of the reduction/smoothing of my decay times comes from room treatment - lots of it. I'm not sure how DSP effects decay times.

Another thing you can do is EQ to taste, as you've already done with the bass boost. For example you may find percussion is not enough, or too much. I found altering 8kHz over a four octave range did the trick (so for example 8Khz is increased say 4dB and 6 and 10 kHz have no change so you get a kind of inversed 'V' shape added to the FR. It's like using a treble tone control. That's the best way to alter the frequencies above the bass region. It still gives you a smooth FR).