• Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Upmixing - where are we at? Have people compared upmixers?

@sfogg - Hope you're well, I suppose by now you've had a little time to play and experiment with your setup. Would you say it is better than L7 via the MC-12? How much better is it considering that the QLI-32 is stem based and provides the benefit of height and overhead channels. Do you feel that the steering algos of the L7 still do a fantastic job?
I've been very happy with DPLIIx via a 7.1 configuration but looking at L7 via a 7.4 configuration.
 
For 2 channel music, yes I think the QLI-32 in alternative mode is overall better than L7 on the MC-12. Occasionally, I find something that is a little too much in QLI and I wish it had the ability to tweak the processing 'aesthetic engine' a little. In that case the flexibility of L7 on the MC-12 was always very nice as you can tune L7 quite a bit. But that situation is pretty rare on 2 channel music and when it does act up it is a weird mix listening in straight 2 channel two. Like nothing in phase between L/R.

There is a fair difference in the envelopment between QLI-32 and L7 on music due to L7 being like a donut of sound around you compared to QLI adding the stage height speakers, the 4 height channels and the additional surround/rear speakers. It is just a more seamless effect.

Ditto for movies, 2 channel or multichannel.

Multichannel music is more of a mixed bag as the end result varies wildly by the mix. Which kind of makes sense in how the QLI processing works as it is really breaking down mixes. If the material in the surrounds is 'dry', meaning it has no reverb/hall sound beyond the main component (which might be direct hall sound), QLI-32 in alternative 5.1 or 7.1 mode can be very effective.

However, if the surround material has additional reverb/hall sound in it (sometimes the hall sound has reverb) then QLI processing is too much as it pulls the reverb out and redistributes that around the room. That can make everything sound too echoey. Ditto if something is steered through the surround channels. If that is 'dry' it is fine. If whatever is steered also has reverb in it then you get this sort of effect where the main sound moves around the room and the reverb is sort of a mirror image of its movement. Which kind of makes sense (a real object moving around a room would have changing reflections too) but it can sometimes be a bit much. I think I mentioned earlier but the organ moving around at that start of the Apple Music ATMOS version of Baba O'Riely by The Who just becomes this confusing mess of reverb as the organ has lots of reverb in the surrounds which gets broken down by QLI and redistributed.

This is where more control over the aesthetic engine of QLI processing would probably be really helpful as I feel like the QLI-32 was tuned pretty aggressively in the 5.1/7.1 modes. Being able to tweak that would likely make this work better for multichannel music but that isn't an option in the QLI-32. So when I run into mixes like that I just listen in bypass mode which plays the MC source straight through.

5.1 Logic 7 in the MC-12 is much less aggressive. It basically steers more to the center from L/R (great for MC music that doesn't use the center well) and steers more surround info out of L/R. 5.1 L7 won't steer things from the surrounds to the fronts so you don't have situations like I detailed above. So for multichannel music 5.1 L7 works better overall. But realistically with the Lexicon I tended to prefer 2 channel sources expanded out to surround compared to MC music mixes anyway as it is typically more enveloping in a natural way. With the QLI-32 I have found pretty much the same thing. QLI processing on 2 channel is more enveloping and natural than the MC mix.
 
Last edited:
Thank you for the detailed explanation @sfogg, very interesting indeed. Do the developers of the QLI-32 keep improving the codec in the form of firmware updates?
 
Thank you for the detailed explanation @sfogg, very interesting indeed. Do the developers of the QLI-32 keep improving the codec in the form of firmware updates?
Nope, last firmware update was from 2016. Harman has amazing tech in QLI processing but has done literally nothing with it for the home market, which is a shame.
 
Nope, last firmware update was from 2016. Harman has amazing tech in QLI processing but has done literally nothing with it for the home market, which is a shame.
They own the tech, but did they get rid of the development team around 2016?!

My suspicion is that the lack of ongoing development, is that they fired the dev team... perhaps because the beancounters associated them with the Harman range of AVR's which was discontinued.

There was also a notable lack of development on the Arcam platforms (both Arcam and JBL branded) possibly for the same reasons

If they hand the QLI code over to the D&M dev team, maybe we will see something interesting come of it.... I somehow doubt that will happen though - how many people are sufficiently interested to pay extra for a QLI feature when they already have Dolby, DTS, and Auro modes to choose from?!

The cognoscenti of 2 channel upmixing are a rare and exclusive group.... realistically, there is no market for it.
 
I was playing some 2ch music on Apple Music tonight thru my new Marantz AV-20.
The latest Dolby Surround with Center Spread active was giving me some of the best upmixing I've ever experienced.
Some amazing individual instrument separation to the rear channels was being heard.
YMMV
 
Nope, last firmware update was from 2016. Harman has amazing tech in QLI processing but has done literally nothing with it for the home market, which is a shame.

You are right indeed, I think a codec like L7 should have been donated to the open source community instead of just killing it. All the effort that David Griesinger put into it should not have gone to waste. The same is the case with DPLIIx.
 
They own the tech, but did they get rid of the development team around 2016?!

My suspicion is that the lack of ongoing development, is that they fired the dev team... perhaps because the beancounters associated them with the Harman range of AVR's which was discontinued.

There was also a notable lack of development on the Arcam platforms (both Arcam and JBL branded) possibly for the same reasons

If they hand the QLI code over to the D&M dev team, maybe we will see something interesting come of it.... I somehow doubt that will happen though - how many people are sufficiently interested to pay extra for a QLI feature when they already have Dolby, DTS, and Auro modes to choose from?!

The cognoscenti of 2 channel upmixing are a rare and exclusive group.... realistically, there is no market for it.

You've made some good and valid points. I think good tech backed by years of R&D should never be discarded. Donate it to the open source community and let people continue to enjoy it as a gesture of good will and giving back to community.
 
They own the tech, but did they get rid of the development team around 2016?!

My suspicion is that the lack of ongoing development, is that they fired the dev team... perhaps because the beancounters associated them with the Harman range of AVR's which was discontinued.

There was also a notable lack of development on the Arcam platforms (both Arcam and JBL branded) possibly for the same reasons

If they hand the QLI code over to the D&M dev team, maybe we will see something interesting come of it.... I somehow doubt that will happen though - how many people are sufficiently interested to pay extra for a QLI feature when they already have Dolby, DTS, and Auro modes to choose from?!

The cognoscenti of 2 channel upmixing are a rare and exclusive group.... realistically, there is no market for it.
I'm not sure if they own the tech or have just licensed it. Check out: https://camdenlabs.com/

I sent them an email as a longshot to see if there was any way of adjusting the aesthetic engine in the QLI-32. Not idea about what happened with the development team. I think the original Lexicon team (Squid Team) was pretty much gone once Lexicon was moved from Bedford. That is why the later Lexicon processors were much more dumbed down compared to prime Lexicon stuff. Don't know who did the work in the QLI-32.

Agree that the 2 channel surround market it small in the home. But expanding movies out beyond 5/7 isn't as small. 2 channel QLI is in the mobile audio market at least somewhat too.
 
I was playing some 2ch music on Apple Music tonight thru my new Marantz AV-20.
The latest Dolby Surround with Center Spread active was giving me some of the best upmixing I've ever experienced.
Some amazing individual instrument separation to the rear channels was being heard.
YMMV

I haven't heard the latest Dolby Surround.

QLI can be downright spooky with how it handles multiple objects panning all over the room in different directions at the same time. (Open Pipe Symphony Track #16 for example) but where it is really amazing is how seamlessly it integrates all the different channels in to such a stable and cohesive whole and the effect that has on the feel of the acoustic space and how much more depth and detail it brings out of music.
 
DSU is OK, but the older DPLII Music could achieve more vivid surround effects. It had a range of Center spreads to choose from, and two other use-adjustable parameters, Dimension and Panorama, that are completely absent from DSU, along with different modes for Games vs Music vs Movie.

DSU seems more aimed at two-channel video soundtracks (which is probably why its default Center spread setting is 'off' -- that emphasizes dialog).
 
DSU is OK, but the older DPLII Music could achieve more vivid surround effects. It had a range of Center spreads to choose from, and two other use-adjustable parameters, Dimension and Panorama, that are completely absent from DSU, along with different modes for Games vs Music vs Movie.

DSU seems more aimed at two-channel video soundtracks (which is probably why its default Center spread setting is 'off' -- that emphasizes dialog).
Center Spread is only useful if your center doesn't match your mains, IMO. It does the opposite of what a good surround system should do; it makes it less discrete by mixing the center channel back into the left/right speakers.

It works best with "center channel speakers" (instead of matching identical mains) as they usually are inferior to the L/R speakers are often placed lower or higher than the L/R speakers. Mixing back into the L/R mains reduces the poorer center sound quality's affect on the mains.

I maintain that if you're the only listener (MLP), you're better off just going to a phantom center (i.e. get the crappy compromised center channel speaker out of the equation altogether!). If you have guests sitting off-center, all Center Spread will do is make the center move closer to the nearest speaker they're sitting closer to (bad bad effect). This is why having 3 identical speakers sounds the best and where Center Spread should always be left OFF.


Dimension is a balance mode that moves the balance front-to-back. You can do that manually with DSU if it means so much to you. One of the complaints about the bass "loudness" equivalent mode in Audyssey is that it moved the imaging front-to-back automatically by loudness.


Panorama moves the out-of-phase sound field further into the back of the room (moves mix of out-of-phase material more to the back of the room. This is not entirely unlike Dimension, but more for the side-to-back effect where Dimension moves the entire soundfield front-to-back (including the front material, which starts to play through the sides and rears).

Overall, they're useful for tuning imperfect room setups or mismatched speakers in some cases (like my Carver Ribbons versus Klipsch Surrounds in my music room that wasn't really set up for surround originally), but not really a different effect, akin to the tuning parameters of Logic 7 on the Lexicon versus the newer HD more limited settings. The less you need the tuning, the less it matters overall. Frankly, I don't find DSU (with overheads turned off) to sound terribly different from PLIIx in most material. Neither compare as well to Neural X for movies or Logic 7 for music, IMO. YMMV.
 
It works best with "center channel speakers" (instead of matching identical mains) as they usually are inferior to the L/R speakers are often placed lower or higher than the L/R speakers.
Well in the reality of the big picture, that's a good thing. I would venture a guess that the number of people who own HT systems that use identical center speakers placed behind something like a sound transparent projection screen is small, maybe on the order of 10%?
Like for many things/times in life, I believe we may looking at some things thru rose colored glasses. The old systems weren't always better.
 
Center Spread is only useful if your center doesn't match your mains, IMO. It does the opposite of what a good surround system should do; it makes it less discrete by mixing the center channel back into the left/right speakers.
I think there is some rationalisation for using it, even if the centre speaker is excellent and identical to the left and right front speakers.

When mixing to 2 channel, the phantom centre cannot be made quite as pinpoint as when 100% of the content is coming from a centre speaker. But the sound engineers were happy with the size and weight of the centre image that they were getting in a phantom centre situation, and probably did not mean to give the listener a relatively pinpoint and diminutive centre image.

Furthermore, when left and right speakers are being used to create a phantom centre, a fair amount of their content is spreading out to the side walls and coming to the listener with additional apparent source width, and this further creates some sense of size weight and breadth to the phantom centre image. When you put 100% of that content to a centre speaker, even a perfect and perfectly matched centre speaker, it doesn't have the benefit of the near side wall reflections, which Toole has shown are positively perceived in stereo.

In summary, I think it is not unusual at all for listeners to try up mixing their music from two channel to multi channel and experience disappointment with the loss of size and weight in the centre image. And I'm sure that the sound engineers did not intended to sound that way, when mixing to 2 channel. Thus IMHO there is a genuine legitimacy to using centre spread in order to get a more preferred centre imaging result when up mixing from two channel.

Cheers.
 
When mixing to 2 channel, the phantom centre cannot be made quite as pinpoint as when 100% of the content is coming from a centre speaker. But the sound engineers were happy with the size and weight of the centre image that they were getting in a phantom centre situation, and probably did not mean to give the listener a relatively pinpoint and diminutive centre image.
Using a center channel in QLI, L7 or Trifield does not give a relatively diminutive center image. Dolby PLII does that if you set center width <3 but that is an artifact of that processing. With QLI,L7 (esp. MC-4/8/12 versions) or Trifield the center solidifies imaging, makes vocal sound more lifelike and in the case of QLI alternative also adds an additional sense of layering on the soundstage which even holds when you move off center. I *think* that is the result of first reflections coming from stage height L/C/Rs but not positive on that.


Furthermore, when left and right speakers are being used to create a phantom centre, a fair amount of their content is spreading out to the side walls and coming to the listener with additional apparent source width, and this further creates some sense of size weight and breadth to the phantom centre image. When you put 100% of that content to a centre speaker, even a perfect and perfectly matched centre speaker, it doesn't have the benefit of the near side wall reflections, which Toole has shown are positively perceived in stereo.

The side wall reflection can add envelopement in stereo. Which the surrounds/rears and potentially heights can do more effectively when upmixing.



In summary, I think it is not unusual at all for listeners to try up mixing their music from two channel to multi channel and experience disappointment with the loss of size and weight in the centre image.

Sure, if you don't have good processing built for music. It is also a shift from how 2 channel sounds and can take a little bit of time to get used to the loss of comb filtering between L/R.
 
Have to say that I think that the processing utilitised by the various upmixers is primarily subjective. I also suspect that ones taste in upmixer, may be partially dictated by the type/genre of music being listened to. I wonder if the fans of L7 here are listening to more traditional genres (rock, classical, jazz), with a more forward/frontal presentation?

My personal favourite upmixer DTS NEO: X definitely works well with modern/electronic music, where it seems more panning and use of the surround channels seems to work very well. However, perhaps it's not as well suited to a more "traditional", L/C/R weighted presentation.
I guess we are blessed to have the options and technology we do, these days.
 
Last edited:
Well in the reality of the big picture, that's a good thing. I would venture a guess that the number of people who own HT systems that use identical center speakers placed behind something like a sound transparent projection screen is small, maybe on the order of 10%?
Like for many things/times in life, I believe we may looking at some things thru rose colored glasses. The old systems weren't always better.
I didn't say anything about the speaker having to be behind an acoustically transparent screen. Having all three be just below the screen, for example is certainly no worse than these ridiculously awful setups with "center channel speakers" sitting on the flipping FLOOR pointed upward because the owner has to have screen at eye level (when ALL theaters up until the 1990s when stadium seating started to appear, were above eye level and most reclining seats shift your view upward anyway). Then there's "dialog lift" (standard feature in Yamaha AVRS, but can be simulated with Mini-DSP or even a mixer in others).

The front (not quite half) of my room is shown here lit up. The center speaker is identical to the left/right (they just have grey grills because I bought them from two different sellers, trying to match my original driver PSB B15 speakers with the identical matching T45 models that have the same setup but with an extra bass driver on the bottom). Other than the Front Heights being slightly offset due to the non-transparent screen (there's a blackout drape/window behind it so no room to do transparent), they're all identical drivers with dialog lift making the entire front stage (not just the dialog) come from about 35% up the height of the screen (Wides are now raised to match that height level; this is an older picture).

Speaking of Toole, he's a huge fan of using acoustic controls that are part of the room. He did something similar in his living room. The Thomas Kincaid art is actually very thick sound absorbing "tapestries" that largely kill the first reflection you're talking about. It was utterly detrimental to the sound. It's so much clearer now and accurate imaging wise. Thick carpeting kills any floor reflections. All overheads are tilted downward towards the MLP (Front Heights are on Pro downward tilt stands with sound absorbing material behind the speakers.

20230531_232739.jpg


Thick blackout drapes serve dual functions and are in the back of the room behind the rear speakers, the front of the room behind the screen and on the left hand side of the room across from the outboard fireplace (it and the bookshelves act as natural diffusers while the drape opposite them kills the return wave). The thick seats themselves act as natural absorption as well. There are both Surround Height and Heights + Tops (10 total overhead) speakers.

Surround Height are only used with Auro-3D (I can use Rear Height, Surround Height or the Auro-3D preferred array of both). All surround speakers are arrayed to Auro-3D and Lexicon (THX Mode) traditional theater style surround arrays (4 sets of arrays, equal to many real theaters of the time period). They also work with Logic 7 due to the extraction processors for 11-channel Logic 7 (25 speakers total including 4 optional floor arrays extracted from out-of-phase ambience of the Mains + Side Surrounds).

20230601_003154.jpg



Here's the RT60 (Studio Level deadened)

RT60 Graph.png


I think there is some rationalisation for using it, even if the centre speaker is excellent and identical to the left and right front speakers.

When mixing to 2 channel, the phantom centre cannot be made quite as pinpoint as when 100% of the content is coming from a centre speaker. But the sound engineers were happy with the size and weight of the centre image that they were getting in a phantom centre situation, and probably did not mean to give the listener a relatively pinpoint and diminutive centre image.

Furthermore, when left and right speakers are being used to create a phantom centre, a fair amount of their content is spreading out to the side walls and coming to the listener with additional apparent source width, and this further creates some sense of size weight and breadth to the phantom centre image. When you put 100% of that content to a centre speaker, even a perfect and perfectly matched centre speaker, it doesn't have the benefit of the near side wall reflections, which Toole has shown are positively perceived in stereo.

The accuracy and "pinpoint" imaging of speakers varies greatly. Any quality speaker will have VERY pinpoint imaging in stereo. Garbage Out = Garbage In (to your ears). Stop using substandard poor imaging speakers and your point is lost entirely.

As for Toole, he was talking about stereo playback only. We're talking about a home theater setup. If you use a PROPER dampened/dead room, there are very few sidewall reflections to "bounce" off of. So what do you do, then? You use a stereo upmixer like Sonic Holography or AuroMatic. DSU/PLIIx/Logic 7 also work great.

These use lots of simulated reflections from actual speakers around that dead (not live) room. In fact, you can simulate any room in existence with such a setup because your room doesn't really come into play (or barely). Do you want to hear the room on the recording or do you want to hear your (inferior) room? I don't want to hear my rectangular room playing live music! I want to hear Carnegie Hall or whatever venue the recording was made at. If you're bouncing sounds off the side-walls, you're hearing your room not the original venue! That is unacceptable.

AuroMatic simulates a decent reflective room for studio recordings (that don't have a venue) so you still miss nothing. In fact, this is where Sonic Holography and Logic 7 really shine by recreating the spatial effects around you like headphones brought out into the room itself.

In this environment and with quality 3 identical pin-point imaging speakers, Center Spread can only make the sound quality worse. I've verified this in testing. They actually sound identical in practice once properly aligned. All Center Spread does is lose the center position LOCK for off-center seating. While I don't have a Center Height speaker (no room), you still get 2x better center positioning with Dialog Lift and Front Height than no center at all (or center spread).
 
Have to say that I think that the processing utilitised by the various upmixers is primarily subjective. I also suspect that ones taste in upmixer, may be partially dictated by the type/genre of music being listened to. I wonder if the fans of L7 here are listening to more traditional genres (rock, classical, jazz), with a more forward/frontal presentation?

My personal favourite upmixer DTS NEO: X definitely works well with modern/electronic music, where it seems more panning and use of the surround channels seems to work very well. However, perhaps it's not as well suited to a more "traditional", L/C/R weighted presentation.
I guess we are blessed to have the options and technology we do, these days.

L7 does what the mix essentially tells it to do. If the music is mixed more forward with less discrete out of phase material that is what you get on playback. If the mix has more things out of phase you get discrete activity in the surrounds.

L7 was also very well regarded for playback of Dolby Surround encoded 2 channel movies. There are many example of material where it would pan sound from right front to right side to right rear to left rear to left side and then left front. It does this even though Dolby Surround encoding was only 4 channels, L/C/R and surround.

Or play a live recording with audience noise and the audience is all around you with the performers up front on stage.

L7 is also an encode format. The MC12 has a neat trick that you can feed it a multichannel source and have it L7 encode the source down to 2 channels. On playback in L7 it maintains a great degree of the discrete nature of the original including stereo surrounds.

Can’t speak to Neo:X but would say for Neo:6 that perhaps modern/electronic music is maybe more forgiving of processing/steering errors compared to more traditional layouts where the errors are more obvious.
 
Last edited:
My experience with L7 is that it is outstanding at adding ambiance to the surround speakers, giving a "live concert-hall"
type experience. But over the years since I've found I preferred a number of the more modern upmixers that I found better
at emulating a real discrete multich source.
YMMV
 
QLI literally breaks down the source into streams and then further breaks down the streams into direct sound, first reflections and reverb. Then it redistributes that over however many channels it is configured for up to 32 discrete channels, I'm running 17 channels and stereo subs. Essentially, making it object based as it isn't 'steering' like most systems.
 
Back
Top Bottom