Trải nghiệm của Integra DRX-8.4: Âm thanh và hình ảnh sống động

So for DLBC, I remain focused around the sweet spot – let’s use “Tightly focused imaging”:

I followed the 8 “box” measurement positions around the main sweet spot, and added a couple of other locations (A and B) above and below the sweet spot to enhance the sampled “volume”. The side positions are +/-12″, vertically approximately +/- 6″ above and below the central sweet spot:

I added a couple of measurement points A & B above and below sweet spot.

After all 11 measurements are captured, now comes the fun part with DLBC! First, select the Bass Control filter design option instead of standard Dirac Live, and let’s create our subwoofer target frequency response:

As you can see, I’m using the “Control points” option so I can draw out the curve; the idea being to correlate the tonal characteristics somewhat to the Dolby Atmos Music Target Curve. (For reference, here are the DIRAC target curves files I used for my sub and the other speakers.)

The white dotted lines represent the detected range of the speaker’s frequency response. Notice in my example that the lower frequency line represents the limit of the smaller Polk PSW111 while the Paradigm Reference SUB1 can clearly go further down below 20Hz. For my sub curve, I’ll get it to roll off steeply slightly below 20Hz (no need for very low frequency rumble!) and around 100Hz at the high end; all the speakers will cross over well within these limits.

Next, we start looking at the speaker groups that cross over to the subs starting with the front left/right. With DLBC, we can set the crossover point for each symmetric speaker set. (DLBC uses a genetic algorithm to search for the best crossover points, so take heed of what the default is because often it’s already optimal.)

Notice that there are some rooms modes around 30-50Hz that cause significant net dips in the frequency response for both speakers. Let’s use the subwoofers to compensate that and have the crossover point at 70Hz.

As you know, the THX subwoofer crossover standard is 80Hz which is low enough to make localization of the subwoofer(s) unlikely; I like keeping it below this if I can just to make the localization extra unlikely.

Next, here’s the center channel:

Nothing much to say here. Let’s keep the crossover at 70Hz matching the fronts.

Rear surround speaker group is next:

These rear speakers are closer to the listening position so room effect is less prominent. I can push the crossover point down to 60Hz without a problem.

Finally, my two height channels for Atmos content:

These are small bookshelf speakers and we can see the frequency roll-off below 70Hz so let’s put the crossover at 75Hz. While height channels can be full bandwidth, typically they’re small speakers (like in-wall units) and often are only expected to reproduce 100Hz-20kHz.

Alright then, with all the targets defined, let’s have the software go “Calculate” (bottom right button) the filters we need.

This 3-step process looks quite different from standard Dirac Live and can take a bit of time. DLBC is said to be using machine learning to optimize the result including mixed-phase correction and the interaction of the independent subs.

Once that’s done we can double check the simulated responses to make sure there are no egregious issues:

Looks good. We now can save the calculated filters to one of the 3 slots available in the Integra receiver. Again, it takes a bit more time to calculate and upload than DL without Bass Control. I see there’s still a known issue with some receivers like the Integra/Onkyo/Pioneer that takes longer than usual to send the filters over the network (as per Changelog).

Let’s export to Slot 1 of 3 in the Integra. Good to add detailed descriptions for future reference…

After exporting, we can have a peek at some of the saved settings like relative channel levels and relative distances/delays to make sure nothing here looks out of place (note the highlighted powered subs both attenuated compared to the main speakers):

I listened to check that the DLBC setting sounded good. Yup, subjectively sounded quite neutral without extra tweaking. In particular, no issue with abnormally low bass which I noticed in the previous Dirac Live write-up (I don’t know if this might have been a bug fixed in 3.11+ when used with the Integra).

As I did before, let’s do a final verification with REW at the listening sweet-spot to make sure things are indeed looking good, and correlating to the significant improvements I heard. As usual, for the measurements, I prefer plotting with 1/12-octave smoothing as a good balance between removing narrow irregularities while providing adequate details to spot anomalies.

Here’s a comparison of frequency response between room correction off vs. DLBC:

Note the +10dB LFE amplitude which is normal. Also the effect of the steep low-pass filter on that LFE channel.

Indeed, not bad at all straight from DLBC without any additional tweaking!

Here’s another look at the frequency response overlaid with 1/3-octave smoothed 5-channel (minus LFE) average +/- 2dB bands to better see the improvement:

By doing this we can see that indeed DLBC has adjusted the average frequency response to be closer to the intended Dolby Atmos Music Target Curve with attenuation into the high treble. Notice that the target curve is counter to my Paradigm speakers’ tendency to rise above 2.5kHz which I find can be a bit fatiguing (I’ll talk more about this in a future post). I know there are those who advocate against full-frequency correction and recommend limiting to bass frequencies (like 500Hz and below). Perhaps if we’re just using basic parametric IIR EQ without phase correction, this might be a reasonable recommendation. However, I have not heard any issues with Dirac Live or other modern DSP implementations (like Acourate or Audiolense) when applied to the full audible bandwidth. The power is yours to be the tonmeister, audiophiles.

While frequency response is most important and audible when irregular, it’s good to also have a look at the time domain to make sure the correction looks tighter. Improved time domain performance will help with transients and can add to soundstage imaging for example when playing music with phasic cues like stereo QSound-encoded albums that add “3D” imaging when seated in the sweet spot.

Here’s a REW step response comparing the uncorrected vs. DLBC-corrected playback taken at the listening sweet spot:

With DLBC, the time alignment has significantly improved across the main multichannel speakers with a tightly aligned initial rise and decay in the first few milliseconds.

In particular, it’s good to see the concordance between the front L+R channels. This is important when playing typical 2-channel stereo material. As much as I appreciate multichannel, most albums are still regular stereo. Also nice to see the absence of pre-ringing.

So how does this all sound?

I don’t think there’s need to belabor the point after all these years that good DSP room correction helps with sound quality whether it’s with standard stereo playback or multichannel. Those who claim that measurements and sound quality do not correlate are simply wrong and obviously unexperienced, or don’t have Golden Ears*. 😄 In-room measurements can tell us a lot and will allow us to anticipate and compensate for issues that might easily be overlooked if we’re not attentive or play music that accentuate those issues.

(* In this regard, I’m OK if a so-called “audiophile” chooses to be fully immersed in subjectivism, doesn’t know how to run his own measurements, can’t interpret graphs or understand the meaning of certain specifications. But please, let’s not be ridiculous and say things like “measurements are useless“, or measurements are “1000% unimportant and doesn’t matter” 🤣. 

What I have learned in my decades of being in this hobby is to not trust most audiophiles who have strong faith in their own hearing ability to the point of reporting uninsightfully on hearing big differences in things like expensive cables (like this, and this) – especially those who also don’t run their own measurements, can’t interpret graphs or understand specifications! Especially if they look older, and one would hope… are wiser. Sadly, those ears are poor measurement devices, and those brain cells need recalibration.)

Test tracks like “Bubbles” for transient response, “Le temps passé” for deep bass and vocal clarity, “Spanish Harlem” for female vocals, and “Time” for soundstage all sounded as good as I have ever heard in my room, passing the toe-tappin’-music-enjoyment test easily.

As an integrated DSP in the receiver, it’s basically set-and-forget; once you’re done, these correction settings will be used with any audio input. In the Integra there are 3 slots that one could toggle between – for example one slot could be the Dolby Music Target Curve, another could be something like the vintage Spendor SA1 curve, whatever you like.

The room correction sounded great when watching movies (like Deadpool & Wolverine watched recently) in multichannel/Atmos. I recently also revisited the audiophile favorite Come Away With Me (2002) streamed in Atmos, again, excellent sounding recording with improved tonal neutrality with DLBC.

Although a bit late (or early!), maybe check out Norah Jones’ I Dream of Christmas (2021) next December. The multichannel/Atmos version of this album sounds quite good. However, with these Norah Jones multichannel recordings, I wish they had used the center channel to anchor her vocals.

In comparison, Andrea Bocelli’s Duets – 30th Anniversary (2024) does an excellent job incorporating that center channel into the mix for the vocals without going overboard on the surround effects (check out “Quizás, Quizás, Quizás” with Jennifer Lopez, nice!). Speaking of duet albums, the recently released remix of Ray Charles’ Genius Loves Company (2004) sounds excellent – available as lossless TrueHD-Atmos download at Immersive Audio Album. I thought the original 5.1 SACD mix was very good already! I see that this remix/remaster has been nominated in the Grammys 2025 “Best Immersive Audio Album” category.

As usual, once everything is dialed in and calibrated, it’s very easy to tell the difference between well-recorded + well-mixed albums versus poorly done productions based on natural tonality, soundstage quality, and low noise level. Like it or not, a transparent system will make it easier to hear nuances in your music and anomalies will be easier to notice. This is what I (we) want in the “high fidelity” pursuit. I do not want my sound system to put a layer of make-up over the sound so everything sounds “euphonic”. Alas, this means we might need to consciously turn off our analytical minds and ignore the blemishes so as to better enjoy the music.

When you run a poor recording through a true hi-fi system, that’s the price to pay; it’s no different than seeing the obvious deficiencies when playing an old VHS recording into your 4K TV! Seriously guys, I imagine vinyl audiophiles have already learned to do this because they’ve been ignoring the high noise level, crackles, pops, etc. from the LP medium since forever.

A quick aside. While I think multichannel/Atmos represents a highly desirable evolution towards creative freedom and expands the immersive musical experience, clearly many multichannel remixes do not sound good. For example, the other day I was listening to a bit of TLC’s album FanMail (1999). Their hit songs “Unpretty” and “No Scrubs” sounded very poor through the multichannel speaker system with weirdly placed vocals, unnatural treble, and excessive rear-channel usage. Many of these poor-sounding tracks seem like they might have been mixed for 2-channel headphone binauralization, probably without being checked in a properly calibrated mutichannel studio using loudspeakers? 

Other times, I’ve come across multichannel recordings with excess bass as if monitored on small speakers with the bass/sub-bass EQ cranked to compensate. I love sub-bass and appreciate the use of subwoofers in an audiophile system, but not being a bass-head, I definitely do not want too much! Basically flat, down to 20Hz is what I feel is most pleasant with my music.

With multichannel mixes, given the extra complexity and variables, I suspect there are more ways to botch the project’s sound quality than simpler 2-channel stereo productions.

Perhaps with some REW fine-tuning, one could achieve similar performance using standard Dirac Live as with DLBC. However, what I can say with this Integra DRX-8.4 receiver is that the DLBC process resulted in noticeably more accurate bass correction right from the start, whereas I needed to put much more time when using standard Dirac Live in order to tweak the results. Plus, the ability to play with the crossover point for each group of speakers (fronts, center, rears, heights) added an extra level of flexibility for those seriously into fine-tuning. I’d say the extra US$299 to upgrade to Dirac Live Bass Control is certainly very reasonable. Subjectively, I believe DLBC has added an extra audible polish to the coherence of the two subwoofers working together in my room.

I remain impressed by Dirac Live’s software and ease-of-use. I hope they fix that slow export to Integra/Onkyo/Pioneer receivers (currently takes about 5 minutes just to upload the filter data!?). Faster calculations and export would certainly help speed up any necessary iterative fine-tuning process.

With some luck, maybe Dirac Live ART (Active Room Treatment) on the Integra DRX-8.4 in the not-too-distant future? 😉 It’s supposed to even further improve room decay time as per their advertised waterfall graphs:

Alright then. With DLBC calibration done, it’s now time to enjoy the music. Cheers!

——————–

Let’s end with the most important thing… Some music.

I was never a fan of the sound of U2’s album How To Dismantle An Atomic Bomb (2004, DR5 stereo) – over-compressed, harsh, nasty-sounding, sadly in keeping with many pop/rock recordings of the early 2000’s. Well, if you like U2 and that album (or maybe want to give it another listen), the recently re-mixed How To Dismantle An Atomic Bomb (Re-Assemble Edition) (2024 release) could be worth a relisten with quite a number of bonus tracks! While the CD version is marginally less compressed at DR6, the multichannel/Atmos version works out to DR12 and features much cleaner, less “shouty” vocals, less harsh guitars, and of course better soundstage envelopment as the mix doesn’t have to be squashed into just 2 channels. Turn the volume up on some “Vertigo”.