Sound Improvements - Roon EQ and Convolution

What sound improvements would you focus on?


They don’t need to improve the sound. They could just always put “sound changed” in the release notes and let people debate it. I can just imagine it. “Sounds MUCH worse, terrible, going back to x other software” “No no no, sounds much better, your in-spec USB cable is too poor for you to be able to hear properly.” When all that happened is @mike typed those words in the release notes… :wink:


I don’t believe Roon can improve the sound, thats the job of my components.
After the hoopla about DSP, I was hoping there would be some better integration with other tools or documentation so we could actually use it.
Curious what percent actual do use the convolution or parametric eq.

What’s so hard about using eq? It’s pretty self explainable imo but maybe I or others can be of help.

The Roon KB page on the parametric eq may help. There are also a few threads in the Roon Software section about PEQ. It is not immediately clear what the Q control is, but it is a bandwidth control. Lower values mean a broader bandwidth. The best way to understand it is to plug in different values for different filters and watch how the slope of the curve changes.

1 Like

@andybob - thanks for providing that information.

@Nyquist - not so much :wink:

I have read all the step by step setup on REW that was extremely well written. We are fortunate to have such contributors.

Still most said the SQ was not better, Most are still looking for ways to take advantage of this new toy.

Well if you are talking about room/loudspeaker correction with self created convolution filters that’s going to be difficult I’m afraid. I’m a sound engineer and I have been designing loudspeakers for almost 40 years now and I am still learning. Not that it has to take this much time, maybe I’m just a slow learner but I’m afraid you are asking the car manufacturer to learn you how to drive the thing out of the showroon. That’s the main reason why many don’t get very good results. It’s not something you do right in just one afternoon. It takes skills but they can be learned.


I am more positive. I found that learning Acourate was a chore, just because it is a toolkit rather than a specific solution for room correction. But it wasn’t unreasonable, and I do not have any years of experience designing speakers. And the result with convolution was outstanding.

I had used PEQ with REW and was not satisfied.

But at this point, with Roon doing convolution, it is great and I abandon MQA (which doesn’t survive DSP) without a moment’s hesitation.


Did not want to sound negative, you don’t need years of experience but their is a learning curve. At least it seems you have been to a driving school.

Maybe it would get those results. On the other hand, if he typed “improved focus and soundstage depth” they’d go bonkers :wink:

1 Like

Very interesting, but it’s another thread, I’m afraid… Nevertheless, could you tell us a little more about your experience with convolution? Do you own horn speakers? What was your aim with convolution?

I don’t have horn speakers (actually I do have Avantgarde horns in another room, but I rarely use that for music). I use Meridian DSP speakers. They are excellent, and have provision for adapting to the room and placement (close to wall or walls, ceiling height, hard/soft walls, seating height).

But nonetheless, my room introduced resonances. It is a reasonable size room, with a high, sloping ceiling, but the walls are parallel and there are resonances. So I use room correction to fix those. The Acourate-generated convolution worked much better for me than the PEQ-type filters generated by REW. These curves, taken with REW, show the response before and after correction:

I also used it in a smaller room, 15’ square and booklined, one glass wall, with bookshelf speakers (actually sitting in the shelves). The results are equally impressive, although with a smaller room sound waves below 30 Hz don’t fit:
(In both cases, the downward slant is intentional, that’s my target response, “house curve”.)

These frequency response measurements don’t tell the whole story. The Acourate convolution also addresses phase/time alignment. I am no expert, but I believe this is the difference from conventional PEQ.

In any case, I am very pleased with the results.

1 Like

Thank you again for your detailed report. The before-after curves made me curious to give it a try with my equipment here. After days of trial and error with unsatisfying results I finally gave up on it, because the result, diagram-wise, showed the same effects like yours but it sounds terrible to my ears. Something missing… no attack, no dynamics, somehow smoothed like behind a curtain. Well, that‘s pushing it a bit but it indicates the trend.

I have to say that my listening environment is in an open-spaced living hall in the house with stone floor and only little dampig like curtains or carpets. The speakers are far from reflective walls as is the main listening place. So, acoustically very spacey, airy and sometimes too thin. Turns out I better invest in thick carpets than in electronics to end up in a smoother and more studio like sound.

And what software did you use to create the filters? Acourate?

I used the Dirac Stereo version with the measurement mic coming with the DENON X7200W. After the first tests I used a DBX RTA-M from a friend of mine but it did not change things dramatically. Both mics are omnis and the latter came with an individual calibration sheet indicating ±1,5dB between 20Hz-20kHz.
What kind/type of mic did you use?

I used the UMIK-1 sold by MiniDSP.
But I don’t think the mic is the issue.
The software that computes the filters is very important.
If you use Dirac, that means it is the Dirac runtime as well, right?

I used the Roon DSP runtime, doing convolution based on filters designed by Acourate which does an outstanding job.
Before that I had tried Roon’s parametric equalization with filters designed by Dirac, and I got very straight lines too but it sounded terrible.

I think the essential thing is that the audio transfer function is a complex function, in the mathematical sense, it has two components. We can think of it as amplitude and phase, phase being related to time alignment. These amplitude curves show only half of the story. The phase/time alignment is hugely important, this has been shown to me many times since I started with this stuff in the 60s. And time alignment is part of the design process you go through with Acourate, it’s not an automatic push button thing, you keep adjusting parameters until you get a good time behavior.

I know Acourate is very focused on this. I think that REW does a simplistic amplitude-only PEQ alignment, so I wasn’t surprised when it sounded bad. I don’t know anything about Dirac, haven’t looked into it because it is a closed system, people speak highly of it, but I can’t say what could go wrong or how you could fix it.

I’ve been exceedingly happy with Thierry’s HAF work. I’ve spent a number of hours with REW and calibrated mic and while I improved the bass response, I gained nothing in time cohesion and phase alignment he has…

1 Like

I can second the HAF filters improvement is very evident. Bass and vocals sound more natural. Also addressed a channel balance issue that I didn’t even know I had. Well worth it…

Based on the information and comments provided on this thread I took the plunge and had HAF develop filters for my room. I used REW with a UMIK microphone to create the room/speaker response. I sent the files to Thierry at HAF and he created filters for both Roon and HQPlayer. The filters work very well; without a doubt there is a big improvement in the sound. Plus it did not require my having to spend time learning new software. I recommend it.


I’m still enjoying his x-talk filters the best after nearly a year. I’ve made enough component changes and room treatments I should have him tune them up.

1 Like