DTS Neural:X

What DTS Neural:X is

DTS Neural:X is an upmixer. It takes content that was never authored with height information and spreads it across the speakers you own, which makes it the DTS counterpart to the Dolby Surround upmixer. It is separate from the DTS:X codec and object format, which carries purpose-built height and object data (see DTS:X). Denon's manual says Neural:X "can upmix your stereo, 5.1 or 7.1 content to take full advantage of all speakers in your surround sound system." DTS's own consumer page describes Neural:X as offering up to 5.1.4 channels with near-discrete front and rear height and wide imaging.

How Neural:X behaves

DTS's public pages describe Neural:X only in general terms, so what follows comes from listening and measurement. High-Def Digest's Josh Zyber compared three upmixers on a Denon AVR-X8500H using the DTS-HD Master Audio 7.1 track of Spectre. DTS Neural:X sent a great many helicopter sounds to the height speakers, often loudly, while crowd sounds barely rose; it was also the only upmixer with prominent drop-outs where nothing came from the heights. The Dolby Surround upmixer did the opposite, drawing diffuse crowd noise upward and leaving most discrete effects on the floor. Zyber's conclusion: Dolby focuses on ambient sound in the heights, Neural:X pulls effects upward, and Neural:X applies a decided volume boost to the heights.

That boost has a number. In a 2016 test of Dolby Surround against Neural:X in a 7.2.4 room with four KEF in-ceiling height speakers, Michael Palmer measured Neural:X at approximately 2 to 4 dB more aggressive in height speaker level, consistent across the material. He noted that with the ear-level speakers running, the volume difference is much harder to hear, and that he personally lifts height speakers about 2 dB when calibrating. Palmer's material was three Blu-rays played as multichannel PCM plus MLB baseball from a DVR in two-channel PCM, so the stereo case was covered too. In practical terms, if you calibrate the heights so the Dolby Surround upmixer sounds right, expect Neural:X to sit 2 to 4 dB above it, and trim the height speakers down accordingly if the overhead effects pull your attention from the screen. Both testers describe subjective results from specific rooms, not a standard, so treat them as character descriptions.

Neural:X vs Dolby Surround and Auro-Matic

The structural differences show up in channel counts. Zyber found that Neural:X and Auro-3D were each limited to 7.1.4 processing on that receiver, ignoring the top-middle speakers, while the Dolby upmixer used all three pairs of heights in his 7.1.6 layout. In exchange, Neural:X and Auro both produce four distinct height channels, front and rear on each side, whereas the Dolby upmixer decodes height as a single stereo signal spread over each side of the room, with no front/back distinction. Dolby's approach fills more speakers; DTS's resolves front versus back better. Zyber described Auro-3D as copying ground channels up with a little reverb. Compare Dolby Surround upmixer and Auro-3D.

Worked example: what changes in a 5.1.4 or 7.1.4 room

A 5.1 source has 6 channels, and a 5.1.4 room needs 10 channels, so Neural:X must create 4 height channels from nothing but the existing ones. A 7.1 source (8 channels) into a 7.1.4 room (12 channels) also creates 4. A stereo source into a 5.1.4 room has 2 channels feeding 10, so 8 of the 10 are synthesized, which is why stereo results vary the most. The 12-channel total echoes the first consumer DTS:X, which Trinnov reports supported a maximum of 12 simultaneous discrete channels, though DTS:X Pro lifts that limit and uses Neural:X to upmix beds to as many as 30 speakers. On Zyber's receiver, a third pair of height speakers gave Neural:X nothing to feed.

Where Neural:X lives

The Denon AVR-X3800H lists the DTS Neural:X upmixer alongside DTS:X and Auro 3D, with 9.4 channels at 105 W. Denon's AVR-X3600H manual describes the Neural:X mode as using the upmixer to extend various sources to multichannel and recommends front height speakers, and the AVR-X4300H manual lists it in its glossary of terms. Palmer's 2017 update notes that Denon's early-2016 firmware let receivers apply Dolby Surround to DTS codecs and Neural:X to Dolby codecs, so the brand pairing is no longer fixed. In high-end processors, Trinnov says its Altitude 32 plays up to 32 discrete outputs for DTS:X Pro. See DTS:X Pro and DTS Virtual:X, which Denon's manual describes as virtual height and surround processing for rooms where height speakers are not used.

Common confusions

Neural:X is not DTS:X decoding, and it does not add objects to a DTS:X disc, which already carries them. The two appear as separate modes on a receiver. DTS also sells a plug-in family named DTS Neural Surround for live broadcast, streaming and post-production, a separate product, so avoid conflating names. Finally, an upmixer cannot create sound that is not in the source: choosing between Neural:X and Dolby Surround is a matter of taste for films, and the DTS:X vs Dolby Atmos and Atmos upgrade guides cover when real object content matters more. Pair it with the receiver match tool to confirm channel counts. Check the receiver's own manual too, because the number of height pairs that Neural:X will use depends on the receiver, as the 7.1.4 limit on Zyber's X8500H shows.