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

[valerianf]

As we are in 2022, it may be a suitable concept to look for a upmixer able to take a 2ch music source and upmix it to 5.1 or higher.
With a 10 years old AVR I am using DPLII music with center spread activated.
Is there a newer upmixer able to do that in a better way?
Center spread is mandatory.

 

[Soundmixer]

As we are in 2022, it may be a suitable concept to look for a upmixer able to take a 2ch music source and upmix it to 5.1 or higher.
With a 10 years old AVR I am using DPLII music with center spread activated.
Is there a newer upmixer able to do that in a better way?
Center spread is mandatory.

If your reference point is DPLII, then DSU (Dolby Surround Upmixer) would be right up your alley. It has center spread processing and it can certainly be matrixed to more than a 5.1 setup. Also, there is the Auro upmixer, which can do the same thing with music and also has some customization settings as well.

Both IMO are more suitable for upmixing stereo music than Neural X. When it comes to upmixing 5.1 and 7.1 movie soundtracks, Neural X is the optimal upmixer IMO.

 

[tifune]

I really wish people understood how the upmixers actually work before making bold statements and proclamations.

Completely agree, which is why I try to avoid such things. The thing with upmixers, though, is try as I might I absolutely cannot find any thorough guide as to what they're actually doing. Why does Neural:X sort of phase instruments in and out of existence in my surrounds, while Auro/Dolby always have something happening in the surrounds? Why is it sometimes the lead vocal comes out of my rear Dolby, but later in the track the lead vocal only comes out of front L/R?

Those are just the first 2 examples that came to me, if you can provide any links or insight it'd be greatly appreciated. Even upmixing software like Halo isn't particularly specific about what's actually happening, just that audio is "pushed" to different Speakers based on arcs and shelves. Would love to know some hard and fast rules like "mono signals go to center, out of phase goes to rear" etc. I realize the algorithms are intellectual property and will largely remain guarded, but surely there's some high level material out there I simply haven't stumbled upon?

 

[NTK]

Completely agree, which is why I try to avoid such things. The thing with upmixers, though, is try as I might I absolutely cannot find any thorough guide as to what they're actually doing. Why does Neural:X sort of phase instruments in and out of existence in my surrounds, while Auro/Dolby always have something happening in the surrounds? Why is it sometimes the lead vocal comes out of my rear Dolby, but later in the track the lead vocal only comes out of front L/R?

Those are just the first 2 examples that came to me, if you can provide any links or insight it'd be greatly appreciated. Even upmixing software like Halo isn't particularly specific about what's actually happening, just that audio is "pushed" to different Speakers based on arcs and shelves. Would love to know some hard and fast rules like "mono signals go to center, out of phase goes to rear" etc. I realize the algorithms are intellectual property and will largely remain guarded, but surely there's some high level material out there I simply haven't stumbled upon?

Up mixer algorithms can be hugely complex. As an example, here is a presentation by Dr David Griesinger on the old Logic 7.

http://www.davidgriesinger.com/surround_from_stereo2.ppt

 

[Soundmixer]

Completely agree, which is why I try to avoid such things. The thing with upmixers, though, is try as I might I absolutely cannot find any thorough guide as to what they're actually doing. Why does Neural:X sort of phase instruments in and out of existence in my surrounds, while Auro/Dolby always have something happening in the surrounds? Why is it sometimes the lead vocal comes out of my rear Dolby, but later in the track the lead vocal only comes out of front L/R?

I can give you a basic idea based on my testing of DSU and Neural X. This is from my post on a Facebook page.

DSU;

"Dolby surround first takes each pair of channels (front, surround, and rear) and processes them individually. If the source is stereo, it first creates a 5.1/7.1 mix from the stereo source and then further processes this to derive height channels. Each channel pair is divided by frequency bands (20) to allow independent processing of individual signals at the frequency level (highs, mids, and lows). After this, each band is processed in the timing domain; direct sounds with phase differences can be steered into the bed channels, while diffuse/decorrelated sound can be steered into the height channels. This offers extremely good separation of channels and highly accurate placement of sounds on the X and Z-axis"

Neural X;

Neural X works quite a bit differently than DSU, as it works much like an object-oriented encoding/rendering system, rather than a phase and delay ambiance extractor. Unlike DSU, which up steers based on variations in the time and frequency domain, Neural X operates entirely in the spatial domain - it is a spatial remapping algorithm. What does this mean exactly? Neural X decodes height information based on the spatial relationship of sounds in a 2.0,5.1/7.1 mix in the X, Y, and Z-axis using a highly accurate signal modeling (a 3D spatial LUT).

 

[dougi]

On my Yamaha AVR, DTS Neo:6 music is the only one that allows you to set the centre spread, so that is my preferred for music upmixing. Generally Dolby Surround for two channel video, but you do get a lot of bleed into L or R on poor material such as newsreaders with stereo lavellier microphones.

 

[dasdoing]

music is the only one that allows you to set the centre spread

intresting https://www.audioholics.com/audio-technologies/dolby-music

So Dolby is still not able to create a center without reducing stereo seperation?

Trifiled seams to be the only way to create a non destructive center? https://www.stereophile.com/content/upward-mobility-2-channels-surround-page-2

 

[dlaloum]

intresting https://www.audioholics.com/audio-technologies/dolby-music

So Dolby is still not able to create a center without reducing stereo seperation?

Trifiled seams to be the only way to create a non destructive center? https://www.stereophile.com/content/upward-mobility-2-channels-surround-page-2

Apparently Dolby PLII could do it, but only with certain prepro's (Bryston) - it was demoed a number of times with Martin Logan speakers, as I recall.

Given that it was quite variable with PLII, and successful only with certain units - it could not be relied on.

One can hope that DSU does better....

 

[tifune]

Trifiled seams to be the only way to create a non destructive center? https://www.stereophile.com/content/upward-mobility-2-channels-surround-page-2

I'd love to try this, but where can you actually get it for any reasonable cost? Seems like it would be dead simple to integrate with modern AVR processing power, or any modern DSP really

 

[dasdoing]

I'd love to try this, but where can you actually get it for any reasonable cost? Seems like it would be dead simple to integrate with modern AVR processing power, or any modern DSP really

there doesn't seam to be any way other than buying an expensive amp. I would love a software version. most AVRs unfortunatly are all movie focussed, and maybe the license is expensive?

 

[dasdoing]

I'd love to try this, but where can you actually get it for any reasonable cost? Seems like it would be dead simple to integrate with modern AVR processing power, or any modern DSP really

there might be a way: https://www.foobar2000.org/components/view/foo_dsp_trifield
can't test it atm

 

[Andysu]

I really wish people understood how the upmixers actually work before making bold statements and proclamations. We are not in the '70s, PLIIx is gone, Dolby Stereo and the old 4:2:4 matrix - gone, and pictures of old theaters and old equipment are deflectively unuseful.

The old days are gone and technology has changed. The upmixers of yesteryear are gone, so it might be more helpful to learn how the news ones work - and THEN make the bold statements and proclamations.

 

[polmuaddib]

My experince is that Dolby Surround works great with 5.1 music (SACD and DVD-Audio) upmixing to immersive format.
If you have Auro and DS on your AVP/R and by chance have Dire Straits - Brothers In Arms SACD, you can play the title song and switch upmixers.
With Dolby, it somehow extrapolates thunder in the begInning of the song and places it correctly in the height channels.
Auro 3D does nothing of the kind. It just mashes all the sound together into a ball of sound floating in the middle. No separation.
For stereo music, I usually use Dolby. Dolby just mildly expands stereo. It’s not aggressive.
Auro seems like loudness is turned on, comparing to video, like dynamic contrast turned up to maximum. Much more aggressive.
I didn’t use DTS Neural X much.

 

[Soundmixer]

So Dolby is still not able to create a center without reducing stereo seperation?

Not without using the center spread function. It is adjustable, so you can dial in as much spread as like.

 

[krabapple]

the up-mixer is a con. it is merely a matrix as some few would know? it doesn't add anything to immerse elsewhere in the room it just extracts a stereo sound that has been matrix filtered to reduce cross-talk and uses one of... well anyone knows how PLIIx sounded with 4.2.4 matrix source and if anyone took the time to actually listen to the surround channels, with LCRsw switched off muted disconnected

Now why would anyone do that? That's like telling people to listen to the difference file of an mp3 versus source , so "it tells you how bad mp3 is".

Wrong. That conclusion ignores psychoacoustics -- notably, masking effects.

There's no way I'm reading that tome you wrote, your sentences are too run-on for my patience level.



DPLIIx Music mode was my favorite upmixer for 2-->5.1. Its successor DSU seems less satisfying. I would love to see a rigorous technical and listening comparison on the various upmixers available now.

 

[Soundmixer]

With Dolby, it somehow extrapolates thunder in the begInning of the song and places it correctly in the height channels.
Auro 3D does nothing of the kind. It just mashes all the sound together into a ball of sound floating in the middle. No separation.

This was my experience with both as well. I also found that Auro 3D adds bass (up to 10db at the highest setting), rolls off the highest frequencies(about 4-5db at the highest frequencies), and up to 15db of second-order distortion below 35hz. That should make the mid-bass sound fuller than the original stereo content, and give the overall sound a "tube" like character.

 

[Soundmixer]

I would love to see a rigorous technical and listening comparison on the various upmixers available now.

I actually tested all three to see what they were doing using test signals and stereo content. The results were pretty interesting. While I prefer DSU for stereo music, and can totally understand why some like the Auro2D upmixer. It "enhances" the sound of the content with a bass boost (+7db) and a gentle roll-off of the high frequencies (-5 at 20khz) when compared to the test signals and content. The higher the setting, the more the bass increased (up to 3db at the highest setting when compared to the lowest). Auro2D is definitely changing the sound, and it is not subtle. It was always at least 3db's hotter than the original test signal and content, while DSU and Neural X remained with 1 dB of the original test signals and content. Neither DSU nor Neural X ventured as far away from the frequency response and amplitude of the original content as Auro2D did.

 

[dougi]

Not without using the center spread function. It is adjustable, so you can dial in as much spread as like.

It depends on the exactly Dolby version and AVR, I think. On my recent Yamaha AVR it is either "on or off" for Dolby Surround. Previous old Denon PLii logic music it was adjustable.

 

[krabapple]

Center Spread is not adjustable in DSU in any AVR. You can only set it on or off. One of the reasons I miss DPLII.

 

[valerianf]

It is very sad that Central Spread is not adjustable in DSU.
It is one good reason to stay with a DPLII capable AVR.