Room correction for stereo listening (Roon and not only Roon)

[NeedForSpeed73]

Hi everybody, I'm into the project of extending the use of room correction DSP, that I've been using for years for home theater to stereo listening (Roon and maybe also CD). So I know how to use REW, not to its full extent, but I know how to take measures with the Umik-1 and have a basic knowledge of how frequency and impulse response curves work (been tinkering with REW filters, my good old miniDSP 2x4HD and MSO).

Lately (2023, when I finally got an AVR with Dirac Live support) I jumped on the Dirac bandwagon and I’ve been enjoying its results (but a little less its “black box” style workflow).

Now I’d like to apply some room correction to my Roon listenings (and not only, see later) and, after some reading here and elsewhere I’ve restricted my choices to three possibilities:

1. The hard (and expensive) way: Acourate. The software looks like REW on steroids, with a high price tag (almost the price of a miniDSP Flex) and a steep learning curve and, for sure, a bunch of functionalities that I’ll never even scratch.
2. The easy and delegate way: HAF. Basically a human version of Dirac, you just take the measures and then the, surely more prepared than me, people of HAF will make the filters for you; and that’s it, no tinkering, no second try, probably also the need to pay again if you change something that make you need to run the process again (gear changes, room changes, etc.).
3. Finally what looks like a good in-between: Focus Fidelity. This one looks like a good compromise between ease of use and the possibility to have a little more control on the process (which would also be part of the fun).

Now what I’m asking you is, what about the final results?

My goal is to achieve the good results Dirac Live (with Bass Control) is giving me for the home theater system (especially in the bass frequencies area).

But I also know that HT is a different kind of beast and especially integration of the subwoofers (that I don’t like to use for stereo listening) make the use of DSP a must, while for stereo listening the benefits are more limited (and even some "audiophiles" totally discourage that).

So would the filters generated by any of the three aforementioned tools/services be able to match Dirac Live’s room correction level (that I've tried with the trials of Audirvana and Dirac Studio) and which of the three would you suggest for a not complete beginner with some will to tinker (and learn) like me?

P.S.: I’m considering also using the calculated filters for CD listening, putting a RaspberryPi based DSP between the CD transport and the DAC (been reading about Hifiberry "hats" and CamillaDSP).

Thanks in advance for your replies.

 

[Everett T]

Using roon you would need a measurement mic and REW and after taking measurements and using REW's eq function you import that file into roon's DSP and convolution folder.

Outside of Roon, Dirac has a PC version, if you use one at your end point or have a laptop laying around . Others may have options for you.

Edit: it's $1k for Dirac with DLBC

 

[Home Audio Fidelity]

The easy and delegate way: HAF. Basically a human version of Dirac, you just take the measures and then the, surely more prepared than me, people of HAF will make the filters for you; and that’s it, no tinkering, no second try, probably also the need to pay again if you change something that make you need to run the process again (gear changes, room changes, etc.).

Thanks for your reference to my service. I would just like to clarify a few points: my service includes the necessary iterations to refine the filters based on client feedback. I offer a significantly reduced rate for clients who wish to upgrade their system, and finally, I provide solutions not available in the alternatives mentioned, such as crosstalk reduction (like Bacch) or separate processing of direct and reverberated sound

 

[Keith_W]

Can you please clarify your proposed system architecture? Draw a quick diagram and upload it to ASR.

 

[NeedForSpeed73]

Using roon you would need a measurement mic and REW and after taking measurements and using REW's eq function you import that file into roon's DSP and convolution folder.

Outside of Roon, Dirac has a PC version, if you use one at your end point or have a laptop laying around . Others may have options for you.

Edit: it's $1k for Dirac with DLBC

I know. I already own the UMIK-1 since many years ago when I started playing with REW and MSO (Multi Sub Optimizer). I know how expensive the Dirac licenses are, luckily my AVR (Onkyo TX-RZ70) had the basic Live full banc license included and just had to pay the Bass Control upgrade.

 

[NeedForSpeed73]

Can you please clarify your proposed system architecture? Draw a quick diagram and upload it to ASR.

Thanks. Your question made me finally decide to draw a diagram of my (rather complicated) system.

 

[popej]

But I also know that HT is a different kind of beast and especially integration of the subwoofers (that I don’t like to use for stereo listening) make the use of DSP a must, while for stereo listening the benefits are more limited (and even some "audiophiles" totally discourage that).

Your AVR doesn't distinguish between music and movie. It optimizes playback for every sound. Simply connect your sources to AVR and profit from Dirac. Use digital inputs, if possible. For example you can connect Raspberry Pi to HDMI input.

 

[Keith_W]

Thanks. Your question made me finally decide to draw a diagram of my (rather complicated) system.

Good lord, man! You weren't kidding when you said it's complicated! The simplest solution is to simply NOT change any components, do all your DSP in your AVR and be done with it.

The problem with attempting to design your own filters via Acourate, REW, or anything else is your AVR. Filter design software needs to be able to see individual channels before it can measure and create filters for it. When you hook your PC up to your AVR, launch REW and see if you can send a test tone (say, pink noise) through each speaker in your system individually. If it can do that, then you are good to go.

Note that manually designing linear-phase filters with REW is not for the faint hearted. REW is great for designing minimum-phase PEQ's, but if you want linear-phase filters it's an extremely steep learning curve, and very tedious. This is because you need to export measurements to rePhase and reimport them. rePhase is pretty cryptic software - I have difficulty using it, and I know a bit more about DSP than most people.

Even Acourate is easier than this, but I don't think Acourate would be suitable for you. Acourate requires the ADC (of the mic) and DAC output to be on the same device - i.e. it needs an interface and an XLR microphone. You can use a USB mic, but the difficulty goes through the roof. Acourate is both flexible and inflexible at the same time. Flexible in that it can be used to design almost anything you can think of. Inflexible in that it forces you to do things a certain way, e.g. that USB mic limitation. The problem with USB mics is latency jitter because ADC and DAC are on different devices. Every time you take a measurement, the devices are started at different times depending on the scheduler on your OS, so timing measurements are inconsistent. This is compounded by the fact that you have a complicated signal chain.

I don't know enough about Focus Fidelity to tell you whether it is suitable or not.

If I was in your situation, I would simply use REW. Start off with manually designing PEQ's using REW's equalizer. It will get you 90% of the way there, and it is likely you will do a better job than Dirac if you know what you are doing. Then you can start dipping your toes into linear-phase.

 

[Everett T]

If I was in your situation, I would simply use REW. Start off with manually designing PEQ's using REW's equalizer. It will get you 90% of the way there, and it is likely you will do a better job than Dirac if you know what you are doing. Then you can start dipping your toes into linear-phase.

Agreed, he already has the mic and is somewhat familiar REW, so the learning curve will be somewhat minimal. It's an in uni weexpensive way to start, and as you mentioned, may achieve a high percentage of his sound goal.

 

[thcdru2k]

I would focus less on the specific tool and more on the signal chain and what you are trying to optimize. In my experience it helps to keep a single DSP layer rather than stacking correction across Roon, AVR, and external DSP, since that makes timing and phase harder to control.

REW with a small number of PEQ filters, mainly below 200 Hz, can get most of the audible benefit. The bigger factor is usually sub and main integration through crossover behavior, delay, and phase, which you can validate with impulse and group delay.

Linear phase and FIR can be useful but are not necessary to achieve very good results.

If your AVR already has Dirac, the simplest approach is to run everything through it.
REW into a convolution engine like Roon or a DSP device is a solid and flexible option.

 

[NeedForSpeed73]

I would focus less on the specific tool and more on the signal chain and what you are trying to optimize. In my experience it helps to keep a single DSP layer rather than stacking correction across Roon, AVR, and external DSP, since that makes timing and phase harder to control.

REW with a small number of PEQ filters, mainly below 200 Hz, can get most of the audible benefit. The bigger factor is usually sub and main integration through crossover behavior, delay, and phase, which you can validate with impulse and group delay.

Linear phase and FIR can be useful but are not necessary to achieve very good results.

If your AVR already has Dirac, the simplest approach is to run everything through it.
REW into a convolution engine like Roon or a DSP device is a solid and flexible option.

Maybe my (too complicated) diagram wasn't clear: I'm never using the whole system at the same time, the AVR and the stereo DAC work alternatively wether I'm watching movies/streaming/videogames or just listening to stereo music (CD/Roon/Vinyl).
So there's no double DSP layer: when using the AV system, DSP is accomplished by the AVR (with Dirac DLBC) and the stereo DAC is OFF; when I listen to stereo music, the AVR is OFF and I'm using the stereo DAC. Yes I know, I could be using the AVR for stereo too, it's redundant and "a waste of money", but that's how I like it and all this is our hobby, so "a waste of money" in any case. I want to have separated Home Theater and Stereo systems (although they share the front speakers and "part" of the stereo amplifier).
Anyways, I've been using this system (with many upgrades and gear replacements) for years now and I've been doing room eq only for the home theater part (where is mandatory), while I've been listening to stereo without any DSP/EQ going on. Now I'd like to try change that and is the reason why I asked for your opinions here (and I thank you all for them).

 

[NeedForSpeed73]

Good lord, man! You weren't kidding when you said it's complicated! The simplest solution is to simply NOT change any components, do all your DSP in your AVR and be done with it.

Yes I know, as I already wrote answering to @thcdru2k, my system is somewhat redundant, but I like it like this: I prefer to have to Home Theater and Stereo system separated (just alternatively using the front speakers).

The problem with attempting to design your own filters via Acourate, REW, or anything else is your AVR. Filter design software needs to be able to see individual channels before it can measure and create filters for it. When you hook your PC up to your AVR, launch REW and see if you can send a test tone (say, pink noise) through each speaker in your system individually. If it can do that, then you are good to go.



Even Acourate is easier than this, but I don't think Acourate would be suitable for you. Acourate requires the ADC (of the mic) and DAC output to be on the same device - i.e. it needs an interface and an XLR microphone. You can use a USB mic, but the difficulty goes through the roof. Acourate is both flexible and inflexible at the same time. Flexible in that it can be used to design almost anything you can think of. Inflexible in that it forces you to do things a certain way, e.g. that USB mic limitation. The problem with USB mics is latency jitter because ADC and DAC are on different devices. Every time you take a measurement, the devices are started at different times depending on the scheduler on your OS, so timing measurements are inconsistent. This is compounded by the fact that you have a complicated signal chain.

The stereo DAC has USB input, so it can be connected to a PC/Mac and be used with REW (I've already used it with the trial of Dirac Processor), thanks for pointing out the Acourate need for an interface and cumbersome USB mic support; I have the UMIK-1 which is USB and it'd been really disappointing to find out I couldn't be using it AFTER having spent almost 500$/€ for the software license. This rules out Acourate from the possible options.

 

[dualazmak]

"Room" correction??!!

If you would like to have nice/useful suggestions here from many ASR colleagues, I highly recommend you sharing not only the signal path diagram but also photos of your audio gears/layout, photos of your listening room with 3D sizes/dimensions, your listening position therein, as well as sizes and materials of floor, mattress/carpet, walls, ceilings, windows, doors, etc.

By the way, you are cordially invited to see and participate on my hosting thread entitled "Let's share diagrams (and photos) of our total physical audio system and the whole signal path, with a few words and/or links" where you can find many of such understandable-at-a-glance photos and diagrams.

 

[thcdru2k]

Maybe my (too complicated) diagram wasn't clear: I'm never using the whole system at the same time, the AVR and the stereo DAC work alternatively wether I'm watching movies/streaming/videogames or just listening to stereo music (CD/Roon/Vinyl).
So there's no double DSP layer: when using the AV system, DSP is accomplished by the AVR (with Dirac DLBC) and the stereo DAC is OFF; when I listen to stereo music, the AVR is OFF and I'm using the stereo DAC. Yes I know, I could be using the AVR for stereo too, it's redundant and "a waste of money", but that's how I like it and all this is our hobby, so "a waste of money" in any case. I want to have separated Home Theater and Stereo systems (although they share the front speakers and "part" of the stereo amplifier).
Anyways, I've been using this system (with many upgrades and gear replacements) for years now and I've been doing room eq only for the home theater part (where is mandatory), while I've been listening to stereo without any DSP/EQ going on. Now I'd like to try change that and is the reason why I asked for your opinions here (and I thank you all for them).

Honestly I focus more on the 2D spectrogram than frequency response. The goal is minimizing excess bass energy and keeping decay as even as possible, especially through the crossover.
I do not treat subs like speakers. I first balance levels around the crossover so neither sub dominates, just enough to make timing behavior clear. Then I work on getting both subs to sum cleanly using acoustic timing reference and the phase tool as a starting point, refining manually.
I isolate each sub and start with the one showing the most excess energy, adjusting delay stepwise from ~0.05 ms down to 0.01 ms, then repeat for the other. I re-check levels as needed since gain and timing interact.
After that I may apply a few light PEQ cuts below 200 Hz, but I am not trying to force a flat response. I also use a light crossfeed where each sub is fed mostly its associated front channel with a bit of the opposite channel.
I calibrate center and surrounds the same way. I run external amps with the AVR mainly handling decoding, and use WiiM/Pi directly into miniDSP for music to bypass AVR completely.
I do not use FIR. I could try CamillaDSP, but haven’t found the need. Dirac ART seems to target a similar outcome automatically, which is why I focus on the spectrogram.

 

[dualazmak]

Honestly I focus more on the 2D spectrogram than frequency response. The goal is minimizing excess bass energy and keeping decay as even as possible, especially through the crossover.
I do not treat subs like speakers. I first balance levels around the crossover so neither sub dominates, just enough to make timing behavior clear. Then I work on getting both subs to sum cleanly using acoustic timing reference and the phase tool as a starting point, refining manually.
I isolate each sub and start with the one showing the most excess energy, adjusting delay stepwise from ~0.05 ms down to 0.01 ms, then repeat for the other. I re-check levels as needed since gain and timing interact.
After that I may apply a few light PEQ cuts below 200 Hz, but I am not trying to force a flat response. I also use a light crossfeed where each sub is fed mostly its associated front channel with a bit of the opposite channel.
I calibrate center and surrounds the same way. I run external amps with the AVR mainly handling decoding, and use WiiM/Pi directly into miniDSP for music to bypass AVR completely.
I do not use FIR. I could try CamillaDSP, but haven’t found the need. Dirac ART seems to target a similar outcome automatically, which is why I focus on the spectrogram.

Essentially agree with you!

My system is L&R 5-way 10-channel stereo setup (i.e. L&R subwoofers, woofers, midranges, tweeters, super-tweeters, all driven by each of dedicated amplifiers), and my tuning policy and procedures have been essentially similar to yours. I use DSP software "EKIO" in upstream Windows PC doing rather mild XOs (crossovers) and minimal mild EQs, as well as Group-Delay, Relative-Gain, Phases, etc. all optimally tuned (objectively and subjectively) for my ears and brain at listening position in my acoustic room environments.

If you and/or OP would be interested, you can find the latest setup of my PC-DS-based multichannel multi-SP-diver multi-amplifier full-active stereo audio system and the summary of my tuning "historic procedures" in my post #931, #1009 and #1022 on my project thread.

Furthermore, you can find here and here the Hyperlink Index for my lengthy thread where you can overview the progresses of my lengthy project thread.

The essence of my present system can be also found with diagrams and photos in post #1 on another my hosting thread entitled "Let's share diagrams (and photos) of our total physical audio system and the whole signal path, with a few words and/or links".

 

[OCA]

You can generate Roon ready convolution filters in REW in about half an hour (including measurements stage) using this tutorial:

Mastering Digital Room Correction

 

[john61ct]

So if I decide I don't want to keep paying to use Roon as a Player anymore

Is there an inexpensive way to just use it for convolution?