INTERACT FORUM

Please login or register.

Login with username, password and session length
Advanced search  
Pages: [1]   Go Down

Author Topic: JRSS Stereo Upmix Missing Feature Request  (Read 10649 times)

hulkss

  • Galactic Citizen
  • ****
  • Posts: 447
JRSS Stereo Upmix Missing Feature Request
« on: January 10, 2012, 12:21:09 am »

The JRSS up-mix of stereo audio (not video) seems to leave the right and left channels untouched. My experience with other up-mix solutions tells me that the material routed to the center speaker needs to be mostly removed from the right and left channels for good imaging. This really improves seats that are off center.

When using JRSS for video up-mix there seems to be more center focus (at least of dialog frequencies).

Would it be possible to provide some selection or adjustability for the "width" or "center focus" of the front channels when upmixing stereo? This would allow a person to pick or adjust for the best sounding front speaker balance and imaging.
Logged

hulkss

  • Galactic Citizen
  • ****
  • Posts: 447
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #1 on: January 11, 2012, 07:35:03 pm »

Bump.
Anyone else interested in this?
Logged

Matt

  • Administrator
  • Citizen of the Universe
  • *****
  • Posts: 41940
  • Shoes gone again!
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #2 on: January 12, 2012, 09:16:22 am »

JRSS with music intentionally leaves the front mains unchanged when upmixing.  In my experience, this sounds the best and fully utilizes the front mains which are often some of the nicest speakers in a setup.

When playing a video (other than a music video), JRSS upmixing uses center-focused mixing.  This moves information common to the front mains into the center.

This approach isn't currently configurable.  Maybe someday.



Logged
Matt Ashland, JRiver Media Center

BradC

  • World Citizen
  • ***
  • Posts: 207
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #3 on: January 12, 2012, 04:22:29 pm »

Matt,

I found that JRSS affected the bass in music even when subwoofer was set to 'do nothing' and for 'stereo sources only mix to 2.1'. That is I don't wasnt upmixing on stereo, but want 5.1 to 7.1 for movies.
I did not like what this did to music, so had to turn it off completely

Is it possible to set it to upmix movies, but do nothing for music?
Logged

justsomeguy

  • Regular Member
  • Citizen of the Universe
  • *****
  • Posts: 525
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #4 on: January 12, 2012, 05:16:57 pm »

I'd love to have this configurable as well. I actually prefer center-focus on most of my music.
Logged

nwboater

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 1346
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #5 on: January 12, 2012, 08:23:01 pm »


Would it be possible to provide some selection or adjustability for the "width" or "center focus" of the front channels when upmixing stereo? This would allow a person to pick or adjust for the best sounding front speaker balance and imaging.

I too would really like this. We have a good center speaker now and soon will have one that is mostly on a par with our L & R.

Matt pointed out that the way it is is good for those with inferior center speakers. Problem is those of us with capable centers are being penalized. Also we have our fronts on the 'long wall'. Most probably have them on the short wall. So the amount of blending required on the center for music can be significantly different in different systems.

Thanks for considering this.

Rod
Logged

hulkss

  • Galactic Citizen
  • ****
  • Posts: 447
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #6 on: January 12, 2012, 10:37:35 pm »

JRSS with music intentionally leaves the front mains unchanged when upmixing.  In my experience, this sounds the best and fully utilizes the front mains which are often some of the nicest speakers in a setup.

When playing a video (other than a music video), JRSS upmixing uses center-focused mixing.  This moves information common to the front mains into the center.

OK, time to get a little more persuasive:

The following comments do assume you have a proper center speaker. Using the same drivers in the center channel as the right and left speakers is a good approach as it provides a "timbre match" across the sound stage

A three channel sound stage has several big advantages:
Foremost, the center channel anchors the sound imaging and eliminates phantom image shifts that often occur in stereo.
Secondly, the center sweet spot is greatly enlarged. You can move to an off-center seat and not loose the dimension and spatial image of the sound. Center placed sounds are still in the center of the sound stage.
Third, the center speaker is clearer because it does not suffer from comb filtering as occurs with common signals from a stereo pair.

The above benefits are not realized if the right and left speakers are playing the information that is redirected to the center speaker. This practice results in the stereo phantom center image competing with the physical center image reducing clarity and focus. The sound is further degraded by the comb filtering that results in any off-center seats when the same material from right, left, and center arrive at different times to the listeners ears.

Does up mixing have an effect on the bass in music. Yes it can. There are two sets of drivers in the room in different places to create the sound of a phantom image. The resulting in-room frequency response, primarily low frequencies, may be different than the sound from a single central speaker. This issue is minimized if low frequencies are being routed to a subwoofer.

As to fully using the front mains. Yes, some poorly mixed material may collapse to center mono when upmixed. I have found that nearly all stereo recordings do not have this problem if properly upmixed.

There are numerous processes documented in the literature for center channel extraction from stereo. My best experience with upmixing has been with DTS Neural Audio.

How Neural audio works: http://dl.dropbox.com/u/45539942/US7929708.pdf

A couple articles:
http://www.sfxmachine.com/docs/FixingThePhantomCenter.pdf
http://www.sfxmachine.com/docs/FrequencyDomainUpmix.pdf



Logged

SamuelMaki

  • World Citizen
  • ***
  • Posts: 115
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #7 on: January 13, 2012, 04:14:13 am »

JRSS with music intentionally leaves the front mains unchanged when upmixing.  In my experience, this sounds the best and fully utilizes the front mains which are often some of the nicest speakers in a setup.

When playing a video (other than a music video), JRSS upmixing uses center-focused mixing.  This moves information common to the front mains into the center.

This approach isn't currently configurable.  Maybe someday.




I agreed with this :) I donīt need any center channels (have set center to move left and right channels), cause I use Martin Loganīs in front. So I use 2/2/0.1 settings for music and movies and the sound is really good... So, JRiver stereo upmix is working as wanted here :)
Logged
AMD Phemon II N930 Quad 2.0ghz, AMD Radeon HD 5650, 4GB RAM, Windows 8 RP x64

flac.rules

  • Regular Member
  • Citizen of the Universe
  • *****
  • Posts: 1260
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #8 on: January 13, 2012, 10:59:23 am »

Matt,

I found that JRSS affected the bass in music even when subwoofer was set to 'do nothing' and for 'stereo sources only mix to 2.1'. That is I don't wasnt upmixing on stereo, but want 5.1 to 7.1 for movies.
I did not like what this did to music, so had to turn it off completely

Is it possible to set it to upmix movies, but do nothing for music?

I had a similar problem, JRSS added +10dB to the sub-channel when upmixing, setting the subwoofer-channel to a crossover (although it shouldn make a difference when you input tdigital to the receiver) fixed the problem.
Logged

flanger216

  • Member
  • *
  • Posts: 3
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #9 on: February 15, 2012, 11:16:46 am »

OK, time to get a little more persuasive...

Hulkss points are all correct --- proper upmixing does not keep the original channels intact, even for music, as this inherently destabilizes the soundfield. To make matters worse, please groove on the following points:

1.) Dolby Pro Logic II's 'music' mode was actually designed for cars, events, showrooms and other environments that require upmixing to >2 speakers but do not provide a reliable center channel, or do not provide one at all. It was originally included in home-theater receivers because, as was previously noted, center-channel speakers used to be frequency-limited to dialog, making them completely unsuitable for music playback. This is no longer true, however --- even Polk's cheapest 5.1 set has a full-range center speaker frequency-matched to the mains.

2.) A surprising amount of music, particularly from the 70s to the early 80s, was intentionally mixed for LTRT matrices. That's right --- probably at least some of your music (especially if it's a non-remastered CD from the 80s) is already encoded for surround sound. When played back through a proper LTRT circuit, you'll get full 5.1 audio with independent stereo surrounds. And it's a fully intentional effect, mixed and steered by the audio engineer. Some sources, such as Cluster's first album, are pretty astounding when properly decoded, almost comparable to discrete 5.1 audio. At one point, several sound sources are panned around you in 360-degrees with near-perfect channel separation. This, mind you, is from a 1971 recording :o

But guess what --- 'music mode' upmixers completely destroy this effect, precisely because they do not remove the phased frequencies from the original channels. In fact, audio that is steered to either the center or the surrounds should be removed from the mains. For example, if you play the aforementioned Cluster sample through a music-mode upmixer, the sound sources do not pan around you at all; instead, they seem to move from directly to your left, to directly in front of you, to directly to your right, and back again.

3.) The performance of any speaker-driver is inversely proportional to the spectral density of its output. This is one (albeit a minor) reason why subwoofer-crossovers are used, but the principle applies for any set of frequencies. Simply adding frequencies to the center and surround does nothing to lessen the burden on the mains, but it does increase the amplitudinal load on the receiver, theoretically worsening its performance. Conversely, properly phasing-out frequencies and steering them to other channels keeps the power-draw the same (as no frequencies are being duplicated), and also very slightly lessens the load on the mains and very slightly improves their performance.

So upmixing with full channel-steering maintains any soundfield better (even for vanilla stereo sources), enables encoded surround-sound cues in some music, and even slightly improves overall system performance. But funnily enough, even Dolby Pro Logic II doesn't offer any good options for music listening --- the movie-mode adds a hardcoded 10ms delay to the rears and aggressively steers all dialog-frequencies to the center, making it unsuitable for music. DTS:NEO seems to steer as much information to the rears as possible (which I presume is why it wins so many A/B tests ::)), again making it unsuitable for 'unmolested' LTRT decoding. Pro Logic I, of course, only has mono surround. Foobar's 'freesurround' plugin is impressive, but has issues of its own (audible phase-artifacts and frequency-shifts).

If JRSS were to offer a properly-steered music mode (which is essentially just a movie-mode upmixer with no delay in the rears and no frequency-based center-steering), it would only make JRiver that much more wicked awesome.

And while we're on the subject of upmixing ;D, mono sources should be played back in the front left and front right channels, or in the front left / center / front right channels, but never just in the center channel by itself --- which is currently how JRSS processes mono. Again, a simple user-selectable option would be great.
Logged

flac.rules

  • Regular Member
  • Citizen of the Universe
  • *****
  • Posts: 1260
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #10 on: February 15, 2012, 03:35:27 pm »

Why would you want to have no delay to the rears?
Logged

hulkss

  • Galactic Citizen
  • ****
  • Posts: 447
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #11 on: February 15, 2012, 08:57:11 pm »

Well....I am expecting a lot from up mixing so I just went out and bought myself a VST plugin solution from DTS that works great in MC. The sound stage I get from up mixing stereo is truly fantastic. It works like this: http://dl.dropbox.com/u/45539942/US7929708.pdf

And looks like this:

Logged

deepsky4565

  • Regular Member
  • Recent member
  • *
  • Posts: 38
  • Change this by choosing profile
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #12 on: February 16, 2012, 03:34:45 am »

So far, I haven't found anything better than DPLIIx for upmixing. That doesn't mean others may disagree, but for me it's the best. I don't like it much at all in its default configuration, but it is the extra parameters in the music X mode that can make it great. I also use the stereo expansion sometimes to give the upmixer more to work with. I really wish we could utilize the DPLIIx .dll from software players in JRiver. Is there some way to accomplish this? I'd like to mix my multichannel music in with stereo, but right now it doesn't work as I have to output stereo in two channel to allow my processor to do the Dolby upmixing, but that won't work with the multichannel output on HDMI seamlessly.
Logged

Sandy B Ridge

  • Citizen of the Universe
  • *****
  • Posts: 885
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #13 on: February 16, 2012, 04:15:59 am »

I'd like to mix my multichannel music in with stereo, but right now it doesn't work as I have to output stereo in two channel to allow my processor to do the Dolby upmixing, but that won't work with the multichannel output on HDMI seamlessly.
I have a mixture of multichannel audio and 2Ch audio and I can send the appropriate number of channels via HDMI to my amp. I'm using WASAPI which leaves the number of channels unaltered (unless you choose to change things in DSP). So for me 2.0 is seen by the amp as 2.0 (and can apply PLII if I wish) and 5.1 or 7.1 is sent as 5.1 or 7.1.
I haven't messed around extensively with JRiver DSP module, mainly because I'd like it to keep 2.0 Music as 2.0 and let my amp do bass management. However, I'd like it to upmix 2.0 Video (eg TV shows) to 5.1 with something comparable to (or better than!) PLII or DTS:NEO. It didn't seem possible to do this when I tried without manually setting up (and switching between) Zones. It would be really nice if music audio could be output unaltered in this way.

Maybe things have changed recently and I haven't noticed?

SBR
Logged

flanger216

  • Member
  • *
  • Posts: 3
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #14 on: February 16, 2012, 07:52:31 pm »

Why would you want to have no delay to the rears?

A 10ms delay in the rears is necessary for LTRT film content, because it's part of Dolby's Pro Logic specs for both encoding and decoding. However, LTRT music content (much of which predates Pro Logic) actively uses the rears to steer sound sources, so delay would hurt or even destroy proper imaging. In either case, one would still use room-correction delays to correct for differing speaker distances, but the 10ms Pro Logic delay is simply to add expansiveness to the sound and is additional to any other delays in the audio chain.

FWIW, DTS Neural UpMix both correctly unwraps LTRT content and is particularly good at deriving surround-information from vanilla stereo. I know a few other audio engineers who love it for professional upmixing jobs. The thing is, at least when I tried it a few months ago, it was limited to 44.1 & 48khz --- which is both baffling on DTS's part and a bit disappointing. Same thing for DPLII Movie Mode, which is a best-in-class upmixer (besides the hardcoded delays, which aren't a big deal unless you're an anal-retentive audio engineer ;)) --- unfortunately, it seems that all hardware and software implementations are limited to at most 44-96khz, which still leaves all my stereo SACDs and DVDAs cold and unloved.
Logged

hulkss

  • Galactic Citizen
  • ****
  • Posts: 447
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #15 on: February 16, 2012, 08:36:24 pm »

FWIW, DTS Neural UpMix both correctly unwraps LTRT content and is particularly good at deriving surround-information from vanilla stereo. I know a few other audio engineers who love it for professional upmixing jobs. The thing is, at least when I tried it a few months ago, it was limited to 44.1 & 48khz --- which is both baffling on DTS's part and a bit disappointing.

Another thing I have noticed about DTS Neural UpMix is that the rendered channels always play something that sounds acceptable. Other upmix processes sometimes make rear or surround output that could best be described as noise. Yes, the sample rate limit is a fact.
Logged

deepsky4565

  • Regular Member
  • Recent member
  • *
  • Posts: 38
  • Change this by choosing profile
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #16 on: February 17, 2012, 06:45:09 am »

I have a mixture of multichannel audio and 2Ch audio and I can send the appropriate number of channels via HDMI to my amp. I'm using WASAPI which leaves the number of channels unaltered (unless you choose to change things in DSP). So for me 2.0 is seen by the amp as 2.0 (and can apply PLII if I wish) and 5.1 or 7.1 is sent as 5.1 or 7.1.


SBR
Thanks, I found I can use WASAPI - Event style, and HDMI to pass 2 channel to my processor, and 5.1 too. I thought this was great, but two other problems now. CD DTS files play reaally slow, and my 4.0 channel flacs won't play. Any way to get this running right with this setup? If JRSS was as good as dpliix, it would be even better. (to stay on topic)
Logged

flac.rules

  • Regular Member
  • Citizen of the Universe
  • *****
  • Posts: 1260
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #17 on: February 18, 2012, 02:15:55 am »

A 10ms delay in the rears is necessary for LTRT film content, because it's part of Dolby's Pro Logic specs for both encoding and decoding. However, LTRT music content (much of which predates Pro Logic) actively uses the rears to steer sound sources, so delay would hurt or even destroy proper imaging. In either case, one would still use room-correction delays to correct for differing speaker distances, but the 10ms Pro Logic delay is simply to add expansiveness to the sound and is additional to any other delays in the audio chain.

FWIW, DTS Neural UpMix both correctly unwraps LTRT content and is particularly good at deriving surround-information from vanilla stereo. I know a few other audio engineers who love it for professional upmixing jobs. The thing is, at least when I tried it a few months ago, it was limited to 44.1 & 48khz --- which is both baffling on DTS's part and a bit disappointing. Same thing for DPLII Movie Mode, which is a best-in-class upmixer (besides the hardcoded delays, which aren't a big deal unless you're an anal-retentive audio engineer ;)) --- unfortunately, it seems that all hardware and software implementations are limited to at most 44-96khz, which still leaves all my stereo SACDs and DVDAs cold and unloved.
I would guess the was majority of music is mixed for normal stereo. If its not possible to detect stuff mixed for LTRT isn't it better to make the upmixer work best on the former? Delay emulates side wall reflections, and will probably give a better sounding experiment for concentional steareo-mixed music.
Logged

Mikkel

  • Galactic Citizen
  • ****
  • Posts: 321
Re: JRSS Stereo Upmix Missing Feature Request
« Reply #18 on: April 19, 2012, 01:40:19 am »

*bump* (in regard to flanger126 and hulkss' arguments)
Logged

Mikkel

  • Galactic Citizen
  • ****
  • Posts: 321
JRSS Stereo Upmix Missing Featre Request (revived from the dead)
« Reply #19 on: March 03, 2013, 04:01:16 pm »

*bump* x2.

Hi Matt,

I take the liberty to bump this thread again to try and persuade you to change the jriver upmixing method. As hulkss states, regular stereo ofte has/creates a phantom center, which is problematic sound quality wise due to the influence of 1st order reflections from both front speakers. With discrete center channels only the 1st order reflections of that single speaker is involved in influencing the sound at the listening position.

Now, when retaining the full original left/right channel output it is practically the same as before only doubled (I think): now you have a phantom images between both left and center as well as center and right, which means a whole lot more reflections causing colouration etc. of the sound image - which as a counter productive byproduct also narrows down the size of the optimum listening spot to preserve the localization of sound.

This clearly cannot be an audible improvement, hence I hope you will reconsider the current method.

My inspiration for posting this was reading page 121 in:
Toole, F. E. (2008). Sound Reproduction. Loudspears and Rooms. Focal Press: Oxford.

Feel free to direct my attention to any misunderstandings of above. I shall gladly stand corrected if I in any way misinterpreted Toole's findings.


Best regards,
Mikkel
Logged
Pages: [1]   Go Up