Room correction - How to start?

[audiofool1982]

Thinking about incorporating room correction. I have REW on my laptop. I hve a custom music server with Euphony OS where Roon core is installed. I also have a UMIK-1 to measure the freq response. How do I play the REW freq sweep from the laptop to the speakers to the UMIK can measure it?

 

My setup: Music server -> Uptone isoregen -> USB DAC(Lampizator golden atlantic trp) -> Preamp -> Poweramp (monoblocks) -> Speakers.

 

Also will I be able to incorporate the results in roon?

 

Thanks in advance for any advice.

[firedog]

You can use Roon for the convolution - read the Roon KB for instructions on the file formats it can use for correction filters.

Not what you want to hear, but after trying both, I paid for Audiolense. It's much easier learn and to use.

[R1200CL]

https://audiophilestyle.com/profile/8172-mitchco/content/?type=cms_records2&change_section=1

I’m using this one. 
https://www.homeaudiofidelity.com/

 

 https://community.roonlabs.com/t/a-guide-how-to-do-room-correction-and-use-it-in-roon/23800

 

[MrDerrick]

I second Home Audio Fidelity.

 

Visit the site, you can download all the tools for free.

 

Very easy to collect your measurements.

[eternaloptimist]

Can I put in a vote for Audiolense XO? Brilliant software and amazing results in my 2 channel passive system. I am in awe of the author Bernt Rønningsbakk.

 

To get it up and running.....

Must read (and then re-read!): https://audiophilestyle.com/ca/ca-academy/audiolense-digital-loudspeaker-and-room-correction-software-walkthrough-r682/. Thank you mitchco!

 

 

In addition, a few comments from my experience getting Audiolense up and running on my 2 channel passive system:

 

USB measurement mics don't work - I have a UMIK-1 from Dirac 1.0 days. Don't bother trying.

I bought a Dayton Audio EMM-6 measurement mic. Relatively cheap. Calibration file downloaded from the Dayton website. Need an XLR to TR (1/4", 6.5 mm) mic cable.

I can strongly recommend a Focusrite Scarlett interface. I used a 4i4 3rd Gen from my music home studio. (The cheaper 2i2 or "Solo" would also work and I suggest the 2i2). You do need TRS (not TR) to XLR cables to connect to power amps directly or system preamp. Has a great mic preamp with 48v supply (required to provide "phantom power" to the mic) and a solid DAC with ASIO drivers that are "bombproof".

Focusrite Scarlett 4i4 with mic plugged into input 1 (left front), TRS outputs 1 and 2 plugged directly into power amps.

I was extremely careful the volume control (attentuator) on the front panel was low and the routing hadn't been changed in software or bang would go the drivers! Also, ensuring that the output cables are plugged into line output 1 and 2 which are volume controlled by the front panel volume control (and not line 3 and 4 which are line level outputs and can be controlled in software but risky!) The Focusrite Scarlett 2i2 / Solo only has one set of line outputs.

 

When setting up, don't expect the "microphone input monitor" to show anything until actually running a measurement. This caused me a lot of headaches thinking that the microphone input was set up incorrectly. Once I worked this out...fine. (I actually used Room EQ Wizard and the SPL meter tool to check the microphone setup. I wish there was similar in Audiolense.)

Also, the Input channel for the Focusrite 4i4 needs to be set at "0" in the Setup Measurement page, even though it is labelled channel "1" in the Focusrite control software and documentation.

 

Setting up the Target curves is not particularly intuitive. No default curves that I could see, you have to start from scratch. Without the example from mitchco in the above article which I followed as tightly as possible, I would have abandoned ship, I think. His included picture of Dr Floyd Toole's preferred target room curves was extremely useful for experimenting e.g. with bass boost.

 

Exporting filters and importing into different software needs a little more explanation, I think....

You have to go into Settings and set the file type of filters to be created before hitting Save filter. 

For JRiver.... then copy the multiple filter files created (in .cfg form) into a single folder, then zip it up! After which JRiver sees it as a "group" of filter files in the Convolver section of DSP Studio.

For Roon.... Audiolense can export ("save") filters as a zipped single folder, but again, have to set the output in Settings. Before I found this, I worked out that grouping filter files in .wav form into a single folder, then zipping it up was recognised appropriately. I understand that this is fine for 2 channel passive systems but not for more complex setups...

I initially created correction files all the way up to 172.4 and 192 KHz but... from reading the Audiolense forums, it seems that there is little point in going above 88.2 and 96 KHz which are then ?resampled in the convolver as needed.

 

Audiolense does offer a convolver, but... although it is a virtual ASIO sound card, it does not include Windows audio sys drivers (as clarified for this newbie by Bernt). What does this mean? It means that Windows doesn't "see it" in Sound preferences as an output device and other software has to be used to interface between Windows and the Audiolense convolver, which then hosts the filter files and outputs to the DAC... (this is my understanding, happy to be corrected).  I found using Hifi Cable and ASIO Bridge (donationware, as suggested on the Audiolense forum) worked https://shop.vb-audio.com/en/win-apps/19-hifi-cable-asio-bridge.html. Alternatively, just using JRiver's WDM Driver (hosting the convolution filters) worked well as an output for Spotify etc.

 

Anyway, just a few longwinded thoughts! 

Not sure if anyone would be interested in a few setup photos from this newbie? A lot of commentary on the forums is from people with complex setups including digital crossovers etc. I couldn't find an "idiot's guide" for people with simple 2 channel passive speaker systems.

 

[firedog]

2 hours ago, eternaloptimist said:

USB measurement mics don't work - I have a UMIK-1 from Dirac 1.0 days. Don't bother trying.

My UMIK worked fine with my Windows 10 Laptop running Audiolense.

[eternaloptimist]

10 minutes ago, firedog said:

My UMIK worked fine with my Windows 10 Laptop running Audiolense.

Hmmm. 

I couldn't get Settings to work that had the UMIK-1 as the input mic with the Chord ASIO USB as the output.

I would like to edit my previous post, but it seems editing is locked... :D

 

[mitchco]

Yes, if using a USB mic with Audiolense, make sure in the measurement window, Advanced Settings menu that "Use Clock Drift Correction" is enabled. This will compensate for two different clocks and works extremely well. I have a post on the Audiolense forum that compares this to an analog mic and the difference was 0.02ms worst case.

 

If you are using a USB mic like the UMIK-1, and an ASIO capable DAC, then you need to download and install:

https://www.asio4all.org/ Make sure you have installed it with "offline" settings checked on. Then open it up and ensure only the UMIK-1 input "blue pin" is enabled and then the DAC playback output blue pin is enabled. Then make sure all other pins are disabled. In Audiolense, for playback and input devices, just select ASIO4ALL in each dropdown and that should map the the settings made earlier. One can verify be checking speaker connections. If still no sound, then in the Advanced Settings menu, select "use separate play and record streams:" and all shall be well :-)

 

Kind regards,

Mitch

[eternaloptimist]

6 minutes ago, mitchco said:

Yes, if using a USB mic with Audiolense, make sure in the measurement window, Advanced Settings menu that "Use Clock Drift Correction" is enabled. This will compensate for two different clocks and works extremely well. I have a post on the Audiolense forum that compares this to an analog mic and the difference was 0.02ms worst case.

 

If you are using a USB mic like the UMIK-1, and an ASIO capable DAC, then you need to download and install:

https://www.asio4all.org/ Make sure you have installed it with "offline" settings checked on. Then open it up and ensure only the UMIK-1 input "blue pin" is enabled and then the DAC playback output blue pin is enabled. Then make sure all other pins are disabled. In Audiolense, for playback and input devices, just select ASIO4ALL in each dropdown and that should map the the settings made earlier. One can verify be checking speaker connections. If still no sound, then in the Advanced Settings menu, select "use separate play and record streams:" and all shall be well :-)

 

Kind regards,

Mitch

 

Thanks Mitch.

At the risk of sounding a bit cheeky.... I think this highlights why the overall take up of DRC with non-experts is low overall... way too much setting "tweaking" required, particularly in the absence of step by step documentation of the main program. Scares a lot of people off, I think.

 

This is why I ended up using the Focusrite Scarlett 4i4 interface: No requirement to install additional ASIO drivers and configure. No requirement to check "Use Clock Drift Correction". It just works.

 

I really do think there would be increased take up of DRC if a straightforward 2 channel passive setup solution was outlined for Newbies with specified hardware e.g. EMM-6 mic, Focusrite 2i2 interface and a step by step setup guide with photos including how to connect the cables with photos of every single step. Just my 2 cents...

 

BTW, Mitch without your contributions in multiple forums, I would not have had success! many thanks.

 

[mitchco]

@eternaloptimistI don't disagree one bit. I spend the majority of my time troubleshooting measurement setups and as the song goes, nobody knows the trouble I have seen :) 

 

You are talking about systems design and this is what I did with my miniPC DSP offering. Everything is already configured, turn it on, take measurements, use my service or design/generate your own filters, install and enjoy! Guaranteed to work out of the box. No fuss, no muss.

 

The issue is very few people are starting from scratch. They already have a computer or a converter or a measurement mic or any combination thereof. Then it is no longer a system design and what can go wrong will go wrong as there are so many variables to account for. This is the fundamental issue. I have detailed step by step instructions and how to guides in addition to what I have written in articles on AS, complete with screen shots, etc., and even then there are troubles. So many, that I could write pages of the issues folks (and I) have run into. 

 

While it is getting better, taking acoustic measurements, analysing the results, designing a custom FIR or DRC filter for ones speakers/room and installing it in whatever convolver is complex task with many variables and makes the big assumption that one knows what they are doing. I certainly sympathize and am actively working on trying to reduce the complexity.

 

Hope you are enjoying the music!

[eternaloptimist]

Duplicate, please remove.

[eternaloptimist]

Mitch, a very good point about most people not starting from scratch. I have a UMIK-1 from Dirac v1.0 days... that didn't go so well as per earlier post!  I can just imagine the trials you would go through professionally to help people get set up!

 

I do think that Dirac were onto something in the past (v 1.0) with their very user friendly GUI, wizard and the Virtual soundcard+convolver (can't remember what it was called). UMIK-1 strongly recommended and easily available (including here in Australia). It really did help that the program generated a suggested starting target room curve. Unfortunately, it seems that Dirac aren't particularly interested in the PC music server market....

I play piano, synths etc. and have a little home music studio so have some familiarity with DAWs, VST and other plugins. I can sort of see why Dirac make their filters in VST form for the studio market, but cannot understand why they can't also release their filters in more typical filter formats (compatible with Roon, HQ player etc.) that could be VST hosted if needed....  

I did actually manage to get the latest Dirac working but the fiddle required... including having to have the Dirac processor running and actually playing music (!) in order for the measuring section of the software to "connect" and work is crazy. Why on earth did they break the previous v1 workflow model that worked..? The end result didn't sound as good as Audiolense in any case.... and was messier ++.

 

And yes, I am really enjoying the music! :D

 

[ecwl]

I just went through creating convolution filters for my system using Acourate a few weeks ago. On one hand, the results are incredible, much better than what I can achieve using REW and frequency/amplitude correction alone. And I don’t find the process too complicated, thanks to @mitchco’s book and other online resources.

On the other hand, I find the time-domain correction/phase correction part incredibly complicated. I ended up creating 6 different filters and listening to them in my system. As Uli, the creator of Acourate said, sometimes, you can be too aggressive with the ExcessPhaseWindow settings and pre-ringing compensation so even though the computed measurements (step response) look better and better, I found that I’m losing transient accuracy, in terms of drum strikes, clapping of hands. Fortunately, I can listen to my headphones as a reference so that I know I didn’t go overboard. I ended up using Time Further Out’s Unsquare Dance and Copland’s Fanfare of the Common Man performed by Minnesota Orchestra to calibrate the filter to ensure optimal transient reproduction.

To me, I wonder if this is the reason why I have not heard a consistently impressive result with Dirac because as Uli said, perhaps these phase corrections are best performed manually, rather than automated as different speakers and rooms have different issues. It also makes me wonder how Audiolense’s time domain correction works (because it looks like it just has a default setting that it doesn’t really talk about as much but you can alter the settings like you can in Acourate). Since Audiolense XO actually costs more than Acourate (which is not cheap either), I’m not sure if I really want to find out...

[mitchco]

@eternaloptimist in the software world, Dirac is a completely closed source solution... Glad you are enjoying the music!

 

@ecwlDirac has no user controls other than target curve. Thus it is a general purpose solution.

 

Both Acourate and Audiolense XO have full user control over frequency dependent windowing. Both low and high frequency window widths can be independently adjusted for both magnitude and excess phase correction, along with the amount of amplitude correction applied.

 

The physics is understanding frequency dependant windowing (FDW) and how that applies to ones speakers and room. There are a number of math calculations to make relative to how much amplitude and excess phase windowing is applied for any given speaker in any given room. Usually at low frequencies one wants quite a long window to take care of room modes/reflections and then above the rooms transition frequency, a lot less excess phase correction, but still some direct sound correction if the speaker is not the smoothest.

 

As pointed out in this article, there is an "ideal response" or textbook response one wants to shoot for the most accurate sound. Of course ones ears count :-)