Dirac Live 3: A look & listen with the Integra DRX-8.4 - the process, microphone placement suggestion, tweaking, and multichannel/Atmos results.

As discussed years ago with my sound room set-up, I believe room correction DSP is very important and has become part of my routine whenever I make significant changes in room layout or when I incorporate new equipment. I would argue that an audiophile who has not seriously spent time trying this likely has not heard the best sound in their room from their system. Simply put, you'll notice a level of control and audible difference way more than just swapping hi-fi DACs, amps, etc. hardware.

With the arrival of the Integra DRX-8.4 receiver recently, I started using the included Dirac Live Full Bandwidth license embedded in the hardware. As I had done in the recent past with Acourate and Audiolense XO, let's dive into the steps I used to get Dirac running and the results I achieved. Hopefully this discussion could be useful for those of you thinking of trying out Dirac.

I'll document the steps I followed, issues I ran into, and tips to overcome problems.

Step 1. Get your room ready.

For my situation, this was documented previously in: The Soundroom - speaker layout, vibration control & correction. Obviously we'll be using Dirac instead of Acourate discussed in that post, and this will be multichannel correction rather than 2-channel stereo even though 2-channel Dirac would still be great.

Other than a couple GIK Acoustics side panels at the first reflection points mainly to dampen vibrations off the glass side cabinets, I haven't really don't anything fancy to the room. Here's a picture of the front equipment set-up these days with the Integra receiver front-and-center on the ground due to size:

Notice that I've got two subwoofers up at the front - the massive Paradigm Signature SUB1 (connected with XLR to Integra) easily capable of pushing frequencies below 20Hz and the much smaller and lower-powered Polk Audio PSW111 (RCA connection) with usable frequencies down to about 30Hz (I discussed the use of 2 subs to reduce nulls awhile back). Based on some previous testing already to ensure that there wasn't a deep null at the sweetspot, the subwoofers were both set to 180° phase, Paradigm SUB1 volume set to 50%, and Polk volume 100%.

The LCR speakers are the Paradigm Signature S8v.3 fronts and matching Signature C3v.3 center channel.

Notice that I already have the measurement MiniDSP UMIK-1 at the listening sweet spot in this picture.

Looking back towards the rear of the room, we have my Paradigm Studio 80v.3 floorstanders used as surrounds and until I have the time and inclination to put holes in my ceiling for front and back heights, I'm just running a couple of Energy C100 bookshelves lifted up and angled up to create the sound of vertical fill. I still have the ELAC Debut 2.0 A4.2 Atmos add-on speakers here but experimenting over time, I prefer the sound of the larger Energy bookshelves with broader frequency response.

Step 2: Get the measurement gear together!

The receiver comes with that little "puck microphone". It's actually pretty good and very easy to use. It has screw-on threads for a camera tripod. Just plug it into the front of the receiver and you can start using that with the Integra Control Pro app (Android, iOS). [There is a YouTube video here comparing the use of the "puck" vs. MiniDSP UMIK-1.]

Since we're audiophiles, we're not afraid of extra work to get things done more precisely 🤓 which in this case means I'll use the MiniDSP UMIK-1 microphone with calibration file. Since we're going to be measuring omnidirectionally with mic pointed upwards, go plug in your UMIK-1's serial number and grab the "<serial num>_90deg.txt" calibration file. A simple mic stand helps position the microphone although I suppose you could just use the UMIK-included tripod assuming you can stably position it where you want (not easy with my leather couch).

I'm running the current Windows Dirac Live v3.10.10 (download here) on the Microsoft Surface laptop in the picture. Be mindful that CPU speed is important as I found that in use when recording the audio sweeps, the old Surface Pro 3 complained of dropping samples here and there so eventually went with my Huawei Matebook X Pro (Intel i7-8550U quad core) laptop instead which provided a much smoother error-free experience.

Step 3: Make sure your receiver is on and the basic speaker set-up is correct.

Using the on-screen menu, make sure the basic layout is correct so Dirac Live knows how many speakers it will be measuring. For me, this would be a 5.1.2 system (while there are 2 subwoofer outs, the system configures them as a single unit):

I let the receiver know that I have both subwoofer 1 and 2 outputs active ("SW 1 & SW2"). And that the single pair of height channels are "top rear" as per my room set-up.

While we're here, Dirac Live requires a network connection so let's just have a peek at the IP address:

As you can see, my Integra DXR-8.4's IP is "192.168.1.107"; make note of this. This IP address can be very important as we'll see below.

Step 4: Hook the UMIK-1 mic up to your laptop, place the microphone at the sweet spot, and fire up Dirac Live.

Alright, connect the UMIK-1 to the laptop, position the microphone at the sweet spot (microphone diaphragm at ear-level as shown in the image above) and run Dirac Live on the laptop. You'll see a screen like this:

Hopefully, the software found your box automatically and you're not seeing that "No devices found." comment like I did!

Normally what should happen is that Dirac Live would look through your network and find the receiver/device with Dirac Live license and proceed to show information about that device with info about the steps ahead. Clearly, the laptop could not find the Integra across my more complex home network with multiple hubs (see latest network diagram here). When this happens, we can follow the Dirac manual connect instructions here.

Since we took down the IP address of the Integra, we now go into the UPnP Descriptor page to have a look for the Dirac Live port number (type "http://<IP address>:8888/upnp_descriptor_0" into your web browser and search for the number):

With the port number 60229 known, we can manually enter this info into the "IP" icon in the Dirac Live opening screen:

With that, you should now see the Dirac Live screen start up with the various microphone inputs including the UMIK-1:

Right click on the UMIK-1 microphone and load in the 90-degree calibration file then "Proceed with Volume Calibration".

As above, we now see all our speakers. You can press the play icon at the bottom of each speaker to hear the relative loudness as "heard" by the UMIK-1 microphone. For safety, the "Master output" is set low to begin with and you'll need to pull it up to hear the white noise from each speaker. With my front left channel at 0dB and "Master output" at -30dB, this correlates with about 80dB SPL using a sound pressure meter (or phone app) which is a reasonable level to be running these tests without going deaf or damaging the drivers:

I'll then press play on each of the channels and adjust the relative volumes so they're about the same. As you can see, I've moved the sliders so each channel is around -25 to -26dB; a value between -20 to -30dB would be fine. Dirac Live will fine-tune the exact relative gain values when it calculates the filters.

Step 5: Select how "focused" you want your measurements.

While in an actual cinema, it's important for the sound to satisfy as many paying customers as possible, for us audiophiles at home, we're almost always aiming to satisfy an audience of one - you sitting in the sweetspot! So I'll select the "Narrow room correction" profile and consider where I want to measure:

As you can see, we can get fully involved and capture data from at all 13 spots around the head and to the sides. While Dirac has those suggested positions shown above, other than the sweet spot measurement, the other locations do not have to be at those places shown.

Since I'm mainly interested in what's happening around the sweet spot, I'm going to do 2 rows of 3 measurements; one row where my ears would be when leaning back on the seat (the Rear row), and a second Front row about 6" forward which is closer to my ear position when I sit up straight.

I've labeled the positions in the image above and here are the actual positions where I would place the microphone around the seating position:

The width of that central seat is 24".

Step 6: Let Dirac Live measure, moving the microphone along those selected measurement locations.

When you click on "Measure selected position" the software will run log sweeps through each speaker in turn, at each microphone position:

Once all positions are measured it'll look like this:

We can see the frequency response curves plotted out for each speaker.

Step 7: Let's define filter targets...

Okay, time to have fun designing the target "house curve". 🙂

These days, since I am listening to a lot of Dolby Atmos-encoded music, let's follow Dolby's recommendation of the Dolby Atmos Music Target Curve as recently discussed. Let me customize each of the curves a bit to conform to that Atmos Music tonality:

I selected the Control Point mode to define the Dolby Music curve. Notice the presence of vertical low-end and high-end "curtain" limits in the graph. You can slide these horizontally to isolate only the frequencies you want to correct. For example, you can pull the right one down to 500Hz if you just want Dirac Live to correct the bass.

Started with the shape of the Dolby Atmos Music Target and manually dropped the upper end of the bass for steeper filter.

Since there's a lot of opportunity to fine-tune, you can always save the project and load back for editing later. Also, you can individually save the curve in one channel group and apply it to another by right clicking on the "Group" box and you'll see load/save menu items for each speaker (eg. center) or group of speakers (eg. fronts, surrounds, heights).

For convenience and reference, here's my full bandwidth Dolby Atmos Music Target Curve file that was used in the images above. You can download other targets here.

Step 8: Let Dirac Live calculate the filters and upload to the receiver.

Click on "Proceed to Filter Export" and select which slot to use. The Integra DRX-8.4 has 3 slots, in this example, let's save to Slot 1 with name and description (as you can see, in this example, I'm replacing a previous variation I was testing):

There are quite a bit of calculations to be done once you click on "Export filter"/"Replace existing filter" so it could take a few minutes depending on the speed of your computer:

If it all worked out, you'll get a confirmation that the filter was exported to the correct slot, and on the Integra DRX-8.4, you can switch between the saved slots (or turn off) using the remote control's dedicated "DIRAC" button:

Make sure to save the project for further editing as needed, and then you can now "Exit" Dirac Live and go have a listen!

Step 9: Fine tuning - subwoofer levels too low and final bass tweaking.

So I select Slot 1 and listen to some music.

Hmmmm... I noticed something odd with the sound. The bass seemed excessively low! This was obvious within seconds of listening to material with strong bass content like Michel Jonasz's "Le temps passé".

[NOTE: It looks like this has been fixed in newer versions of Dirac Live. Worth double checking but I've noticed low bass volume no longer an issue with my Integra since Dirac Live 3.11 and also the latest Integra November 18, 2024 firmware update.]

Let's fire up REW and have a look, here a sweep of the front LCR channels plus the LFE subwoofers:

Interesting. Frequency response looks and sounds good from 70Hz going above the subwoofer crossover, but below that, clearly the bass is dropping off abnormally! It's as if the system is equalizing the LFE level to the rest of the channels rather than allowing the LFE to be at +10dB as normally intended (and subsequently lifting the sub-bass for the other channels).

Let's log into the Integra DRX-8.4's webpage configuration (the device's IP address, default login and password "ciuser") and see what the relative channel levels have been set to:

For some reason Dirac Live seems to be over-attenuating the subwoofer levels, especially the more powerful Subwoofer 1 (Paradigm SUB1). As far as I can tell, I don't think I've turned on other settings that could be interfering with this.

As annotated in the image above, with some trial-and-error using REW measurements to determine how much to add, I increased the subwoofer levels up to +2 and +4dB. As part of the fine-tuning, I loaded the saved project and tweaked the subwoofer target curve shape for better sub-bass consistency, saving this into Slot 2 in the receiver:

Hmmm, maybe I need to target this to +10dB so there's no need to tweak the levels up, as above? Will try this next time...

BTW, just as the Subwoofer 1 level seemed unusually low, I noticed that Dirac Live didn't properly set the distance/delay either:

Strange that there was no value set for the more powerful Subwoofer 1. Unfortunately I don't see a way to manually change this. Am I missing something, or maybe this is intended with this Dirac implementation? The other speakers were all distanced in the 20-30msec range.

Step 10: Enjoy the music! And final verification.

Okay, let's now enjoy some music with Dirac Live Full Bandwidth dialed in. It sounds great now with excellent bass response!

For my edification and to show you my final results, let's verify that objectively the effect has been beneficial.

First, here's the frequency response for all 5.1 channels measured in REW before any kind of bass management or calibration at the listening sweet spot:

Nasty! That's what it looks (and sounds) like if you throw your speakers in a room without taking the time to at least use some kind of calibration system to adjust levels, crossovers, or relative distance of speakers from the listener. It's not unusual to see bass frequencies bounce around a 15-20dB range in most domestic listening rooms. As per Mitch Barnett's previous article on Dirac Live, below the Schroeder Frequency (typically ~200Hz), "the room is in control". Trained listeners should recognize the uneven tonality if not the somewhat "blurred", unfocused transients, on a track like Yosi Horikawa's "Bubbles" with Dolby Surround Upmixer turned on to engage the surround speakers.

Here's my final tweaked Dirac Live frequency response at the listening sweet spot:

Speaker tonality much more consistent, and frequency response tightened to about 10dB range now. Out of curiosity and for visual clarity, let's average the 5 full-bandwidth channels to get a sense of the system tonality with +/- 2dB range:

Other than the occasional dip, mainly in the surround channels, frequency response is within +/-2dB of the average.

There's a small, about +3dB bump from 20-30Hz I could suppress if I wanted to I suppose. That visceral sub-bass intensity can be exciting when the music/movie hits those subterranean registers! Note that this kind of relatively mild bass bump is small compared to the +6dB bass target we often see such as with the Olive & Welti target, intensity typically dipping down by 20Hz.

For reference:

From Olive, Welti & McMullin discussed in Toole (2015).

In the same way we can measure the time-domain step response to see the before-after effect with Dirac Live. Here are the step responses captured over 70ms:

A mess with speaker delays all over the place uncalibrated!

Significant temporal improvement as well. No pre-ringing artifact from the processing. The only issue is that the subwoofer appears a bit off - compare this to the results from Audiolense XO DSP simulation of ideally what the LFE step should look like. With the subs crossed at 70Hz, in practice I don't think I notice the difference even though it should be tighter. Perhaps the subwoofer time-domain performance is a result of the lack of delay set for "Subwoofer 1"?

While we're looking at these room sweeps, we might as well have a peek at the RT60 reverberation time in my room to make sure it looks OK:

Nothing scary here. Not much variation across the audible frequencies and among the different speakers. I'm happy to see RT60 below 0.5s in this room.

One last thing, let's have a peek at where Dirac Live has put the crossover points for each speaker:

Nice. 70Hz across the board including the height channels. The THX standard is 80Hz and having a lower crossover point will further reduce chance of subwoofer localization; which is even less of an issue with the 2 subs.

While not shown here, I also took some measurements 1 and 2 feet on either side of the sweet spot. These looked pretty good as well and similar to the variation discussed before. No, one does not need to sit exactly in the sweet spot to experience the benefits of the DSP correction.

Conclusions:

With Dirac Live, the listening experience has audibly improved. No doubt, it is a powerful tool that will let audiophiles and home theater enthusiasts optimize their sound quality effectively and relatively efficiently even if not completely automatically as in my case.

Like with other room correction DSP packages (eg. Audiolense XO, Acourate), the sound quality became smoother and better controlled. The benefit of using Dirac with the internal Integra DSP is that all speakers, including the height channels are handled in-the-box - no need for external hardware, software plug-ins or activate the convolution feature in Roon. Despite the extra steps needed because Dirac Live did not find the receiver in my home network automatically, the user interface was intuitive and smooth. Make sure you have a reasonably powerful CPU when using a laptop (I would recommend a modern quad-core) to reduce capture errors and improve the time it takes to calculate the filters.

Like with other power DSP filter-creation software, Dirac Live does require some know-how. As much as it would have been nice to just let the software automatically create the filters and expect a good-sounding calibration, in my case, I needed to put a bit more work into it when I realized the subwoofer output levels were low. Furthermore, it was beneficial to use REW to objectively guide further bass tweaking; it would have been impossible to achieve tight results simply "by ear"!

After the tweaking, Michel Jonasz's "Le temps passé" now sounds impressively deep and inviting (the track is off La Fabuleuse Histoire De Mr Swing, 1988, DR15). The dissonance of Sonic Youth's "Superstar" (off If I Were A Carpenter tribute album, 1994, DR9) is rendered with jarring emotional nuances. There's excellent spatial depth on Antonio Carlos Jobim's relaxing Wave (1967, DR10). The improved time-domain performance added to the surround effect of the Q-Sound encoded The Immaculate Collection (1990, DR12) and solidified the illusion of sounds emanating from all around the listener of a modern track like Francine Thirteen's "Queen Mary". And this is all just with 2-channel stereo. Even more to hear from the surround processor when fed DSP-corrected spatial multichannel/Atmos content! 😏

Audiophile companies would like us to think that the "high end" represents state-of-the-art sound quality which somehow over time will trickle down to more affordable products; I think many audiophiles (including myself) would argue that we've already exceeded the point of excellence in many reasonable-priced products and where we need to look is beyond mere 2-channel fidelity. The inclusion and evolution of better room-correction DSP in modern AV receivers is an example of taking a step forward to address room acoustics which are just as important as the sound of speakers and other hardware regardless of what price-point manufacturers are targeting.

Perhaps understandably, there are certain manufacturers who don't like the idea of room correction DSP, rather, they espouse superficial fluff, often coming from olde guys claiming to "hear" things with no evidence. For example, Paul McGowan of PS Audio thinks that the use of Dirac Live somehow creates a situation where one has to "hold your head in a vise" (see here, 1:10) to have the DSP be effective. That is blatantly untrue, especially when we're also taking multi-location measurements into account. Common sense would tell us that if it's that bad, nobody would want to use it in home theater applications! Yet this is where it's being implemented.

Moving one's head away from the targeted sweet spot regardless of the use of DSP is obviously sub-optimal, drop in quality dependent on room modal changes. The only difference is that without DSP, the listener is starting at a lower level of accuracy regardless of "high end" hardware. We must make sure to be mindful of the various parallel domains that work together to create "good sound" - the room itself and the various acoustic "treatments", whether physical room treatments or digital correction can be just as important as any hardware we buy. Hardware manufacturers conveniently downplay or ignore talking about this for the obvious reason that this is not stuff they sell. The wise, insightful audiophile must always keep this in mind.

While the results are excellent already, I am contemplating the US$299 purchase for the Dirac Live Bass Control (DLBC) upgrade. I imagine this will include independent target curves for each subwoofer, presumably no concerns about unusually low bass level, as well as more accurate subwoofer delay settings. For those who are using it, would love to hear your thoughts about DLBC!

--------------------

One last thing, I see that new firmwares have been released or are coming soon for a number of newer Onkyo, Integra, Pioneer (Elite) models. It's suppose to improve Dirac Live measurements, improve Amazon Music hi-res support, Tidal MAX hi-res streaming and both Amazon + Tidal Dolby Atmos streaming.

Hmmm, if multichannel/Atmos streaming works well with Amazon Music or Tidal, I'll have to consider switching over to those services instead of Apple Music. I wonder, is there a count as to how many Atmos albums are available on each of the services (Apple, Tidal, Amazon)?

I'm still hoping that Roon does a better job of supporting multichannel/Atmos in general and increases support for multichannel with Roon Ready devices like this Integra (discussed last time).

Hope you're all enjoying the music.

Addendum:

As I was ending the write-up, I ran across this interesting video about a microphone placement recommendation with Dirac Live:

Interesting that he's using a similar system as what I'm doing. I see he's also running 6 measurements in 2 rows but 8" horizontally and anteriorly rather than my 12" horizontally and 6" anteriorly. I suspect the effect would be about the same and the outcome sounds great. 

I echo his sentiment that the only persons who cares about optimizing sound quality to this degree is ourselves, so might as well focus on the sweet spot where we sit.

He suggested some Atmos material to test out: the car racing scene in Ready Player One (~13:00-15:00), check out the first 15 minutes of Gravity, and the Dolby Atmos Shattered demo. He also suggested the headphone-music-car-chase opening scene from Baby Driver which I haven't listened to yet. For dts:X, check out the "Gilderoy Lockhart" scene in Harry Potter and The Chamber of Secrets at 36:00 with cornish pixies flying around - very cool - and the "Rogue Bludger" (53:00) scene with the Quidditch game.