As part of my testing of the flagship Denon AVR-X8500H A/V receiver, I finally have access to all three of the competing immersive sound upmixer formats and was able to run a comparison to see exactly how they each work on a normal movie soundtrack.
My prior A/V receiver, the 2014 Denon AVR-X5200W, was a Dolby Atmos model released prior to the introduction of DTS:X. In addition to being unable to decode actual DTS:X soundtracks (which defaulted to the DTS-HD Master Audio 7.1 core instead), the receiver also lacked the corresponding DTS Neural:X surround upmixer. Its support for DTS upmixing was limited to Neo:6, which topped out at 7.1 channels of sound. As such, in order to fill the height speakers in my room when playing a non-Atmos soundtrack, I had no choice but to use the Dolby Surround Upmixer (DSU). I didn’t have much complaint with this, as DSU generally does a good job. Still, I’d been curious to see what the competition did differently.
A couple years ago, our site’s Michael S. Palmer made his own comparison between DSU and Neural:X, which suggested that Neural:X was more aggressive and louder in panning sound effects to the height speakers. Whether this is desirable may be a matter of personal opinion. Not only did I want to check that out for myself, the X8500H affords me a couple of advantages that Michael didn’t have at the time: 1) I can run six height speakers in a 7.1.6 configuration rather than be restricted to 7.1.4, and 2) the X8500H also comes standard with the third-party Auro-3D upmixer, which was previously a $199 add-on expense (that I doubt many users paid for).
After installing the X8500H, I configured the receiver for 7.1.6 format, designating my three pairs of height speakers as Front Height, Top Middle, and Rear Height respectively. Then I performed an Audyssey calibration to adjust the volume levels and apply EQ.
Of the three upmixers, only DSU utilizes all three height pairs. Both Neural:X and Auro-3D are limited to 7.1.4 processing, ignoring the Top Middle speakers. (Auro-3D can be configured for 5.1.6, but I did not test that this time through.) On the other hand, although it fills more speakers with sound, DSU only actually decodes height information as a single stereo signal, using the entire left side of the room (all speakers) as one channel and the entire right side of the room as the other, with no distinction between front and rear. (Note that this only applies to DSU upmixing, not to native Dolby Atmos soundtracks, which have fully discrete sound from every speaker.) In their favor, both Neural:X and Auro-3D have four distinct upmixed height channels, front and rear on both the left and right.
My go-to demo for upmixing is the DTS-HD Master Audio 7.1 soundtrack on the Blu-ray edition of ‘Spectre‘. The opening teaser action sequence has a huge crowd of thousands of people on the ground shouting and cheering during the Dia de los Mueros celebration in Mexico City while a helicopter does acrobatic stunts above them. I started by watching the scene with all three upmixers. In all cases, I could hear the sound of the helicopter moving around above my head, which is especially effective when it pans from one side of the room to the other (left to right or vice versa). While there were some differences, they all sounded more similar than not.
To get a better sense of what each upmixer really does, I then disconnected my subwoofer and all of my speakers except the height channels. The results I got after replaying the scene were interesting:
DSU – It was very startling to me just how few of the helicopter sounds emanated from the height speakers in this scene. Almost none, in fact. Those sounds almost entirely remained in the ground level speakers, but they image extremely well, enough to fool me into thinking that the sounds were above my head. Mostly what DSU drew upward were the diffuse noises of the crowd shouting and cheering, which filled the whole room through the entire scene even when only using the height speakers.
DTS Neural:X – This behaved almost exactly the opposite of DSU. A great many helicopter sounds came from the height speakers, often very loudly. Meanwhile, crowd sounds were hardly drawn upwards at all. Logically, this would seem to make sense (the crowd is on the ground and the helicopter is above), but I still expected more of the ambient crowd noises to fill the heights. Neural:X is the only of the three upmixers to have prominent drop-outs where no sound at all came from the height speakers during this scene.
Auro-3D – Auro fell somewhere in between the other two. It had a fair number of both crowd noises and helicopter sounds in the heights. The crowd noises were roughly equivalent to what DSU did but with more helicopter sounds. Those helicopter sounds were not nearly as loud as Neural:X.
I then reconnected my subwoofer and all my other speakers. Regardless of what I learned about how much the height speakers were actually working, the helicopter still seemed to buzz above my head using all three sound formats. It was indeed loudest up there with Neural:X, but it was plenty convincing with the other two as well. As I said earlier, the overall sound of the scene was largely similar regardless of which upmixer was chosen.
It seems to me that each of these three processors has a different philosophy for how to upmix audio from a standard channel-based soundtrack into an immersive system. The Dolby Surround Upmixer primarily focuses on drawing ambient and atmospheric sounds into the height speakers while leaving most discrete sound effects in the original ground channels. DTS Neural:X does the opposite, pulling sound effects upward while leaving ambient sounds below. It also applies a decided volume boost to the height speakers compared to the other formats. Finally, Auro-3D appears to simply copy the ground channels into the heights, expanding the soundstage vertically with a little reverb added to give the heights some separation.
The lack of Top Middle support in Neural:X or Auro-3D didn’t prove to be much of an impediment. Both imaged sound above my head fairly well using just the Front Height and Rear Height channels. At the same time, I hardly noticed that DSU did not have discrete front and rear effects, likely because the types of sounds it pulls upward don’t rely on that kind of precision.
Which of these does a better job of upmixing will be a matter of personal preference, and may also vary depending on the content. With its volume boost and focus on discrete sound effects, Neural:X is perhaps the showiest of the three upmixer formats. It really wants you to know that it’s working. As a result, it can have more of a “wow” factor than the subtler DSU. At the same time, Neural:X risks being too aggressive and loud, which can be distracting. Its focus on sound effects also means that it risks erroneously drawing inappropriate sound effects that are intended to stay at ground level (like, for example, cars driving) above the listener. Auro-3D can also have this problem, whereas DSU takes a more cautious approach and is more likely to leave the sound effects where they started, instead keeping its heights constantly active with atmospheric effects.
Auro-3D essentially creates walls of sound, as if each of the speakers on the ground were taller and extended to the ceiling. This can be quite effective in some circumstances, especially music, but in my opinion is less useful for movies because it cannot provide true separation where some sounds leave the ground and only appear overhead. Conversely, that type of separation can be a detriment for music. When watching a concert, you don’t want the sound of a flute or violin to appear isolated above your head, which can sometimes happen with DSU. In those circumstances, Auro-3D shines.
Ultimately, even despite their differences, all three of these upmixers are pretty good at what they do. With all speakers playing on both the ground and height levels, they’re all liable to have quite similar net effects. The best course of action is to play around with all of them and default to the one that you find most engaging for the broadest variety of content.
Dolby Gets Petty
All of this testing of upmixers may unfortunately be complicated by recent news that Dolby has reportedly mandated that its licensing partners prohibit cross-upmixing of Dolby-encoded soundtracks with competitor processors. In other words, if you watch a movie or other content with a Dolby Digital or Dolby TrueHD soundtrack, you’ll be restricted to only using the Dolby Surround Upmixer, and cannot apply DTS Neural:X or Auro-3D upmixing to it.
There is no technical reason or limitation for this. Cross-upmixing works fine on current A/V receivers. Although certain 2015 processors from Denon and Marantz did block cross-upmixing in this manner for a time, that was later resolved with a firmware update that re-enabled the feature.
Right now, it’s not clear whether this mandate will only apply to upcoming products, or whether owners of existing A/V receivers and processors should worry about currently active features being taken away during firmware updates. Until that’s cleared up, it may be wise to disconnect those units from their internet connections and avoid making firmware updates.
In the event that cross-upmixing is blocked, viewers watching movies on Blu-ray or Ultra HD discs can switch their players to outputting the soundtracks in PCM 5.1 or 7.1 form rather than Bitstream. Unfortunately, this workaround is unlikely to work for TV and cable broadcasts or streaming, where the receiver boxes typically limit PCM output to just basic stereo sound.
To be frank, this seems like a major dick move to me, done for no apparent reason other than to spite the competition. At present, DTS has not imposed a similar limitation on its own soundtracks, but I wouldn’t be surprised to see that company retaliate in kind, blocking DSU from use on Blu-rays with DTS-HD Master Audio sound. In the end, it’s the consumers who will hurt the most from this. I hope the powers-that-be at Dolby rethink this decision.