Old thread for obsolete version of V.I bidule layout

[ursamtl]

Originally posted by kempfand
Preservation_5_Speaker, as well as the "LCR Gerzon" and the "Improvement" Bidule Groups described in LCR Upmix Bidules (http://forum.doom9.org/showthread.php?&threadid=70678&) are based on Gerzon's 1997 work.

For details, see US Patent # 5,594,800: "Surround Reproduction System Having A Matrix Converter". If you are interested in the equations used, you can use the patent number search at US Patent Org (http://patft.uspto.gov/netahtml/srchnum.htm) for "5,594,800".

It is also referenced as "Trifield Decode" (present in some high-end Meridian HW systems).

Cheers,

Andreas

Thanks for the references, Andreas. I've added them to the list.

For those interested in these types of discussions, you might want to join the sursound mailing list, Sursound at https://mail.music.vt.edu/mailman/listinfo/sursound. No need to get individual messages, you can subscribe to their digest as I have for the past few months and read discussions by the leading experts in surround. For this specific topic, a long thread occurred this summer discussing LCR theories put forth decades ago. There's also been an enlightening discussion on DVD-Audio and SACD formats.

[daphy]

Originally posted by krbo
[BFor now I must examine one problem in reproduction, looks some channel are too boosted so clipping is audible.

Test material was from JMJarre, ripped goes straight to 0dB so if there are any amplification in VI module there will be problem.

Must demux bidule output to check each wave. [/B]

Hi all,

I did some testing yesterday and I could confirm this behaviour @ SL/SR too. I used the latest version of V.I (oct. 2004) with default setting (+LFE enabled) with some 32bit 48KHz sources. The 'back side' was totally over boosted - for listening I have to lower SL/SR about -10 db but still they were clipping/scratching :scared: as shown in audition the 'back side' level is a multible from the frontside speakers.
Today I will try it one more time with diffrent settings inside the bidule :o

[ursamtl]

Hi Daphy,

I'm surprised you experienced this problem, especially given your earlier positive results (http://forum.doom9.org/showthread.php?s=&postid=536404#post536404). You might recheck your settings because from the couple of dozen people who've given me feedback, only you and krbo have mentioned this problem.

It might be at least partially due to the source material you used. One trick you can borrow from professional sound engineers is to put a limiter on the output of the tracks if there's a problem. One good (and free :)) choice would be Kjaerhus Audio's Classic Master Limiter available at http://www.kjaerhusaudio.com/classic-series.php. Be sure to set the threshold 0.0dB or -0.5dB. Anything lower than this and you'll be start boosting the overall average level. This can be useful for some purposes, such as to tighten up an LFE channel, but in your case, it would defeat the purpose. However, at 0.0dB or -0.5dB threshold, you're basically brickwall limiting to prevent clipping. If you put this on two surround connections coming out of the Outputs group, and there is still clipping, then it's a problem with your source.

Give me just a few more days to finish some guide writing and you'll have a new version of V.I that is going to be quite interesting. :) Tests we carried out on it this weekend turned out very well.

Steve.

[daphy]

back to the source issue:
I took ~45 min of diffrent material in one file, so there are many diffrent tracks associated in one go with of course diffrent peaks and RMS.
In the moment I still can´t believe the dependency of source to overboost.

One other interesting thing could be the format:

Steve, did you ever test a 32bit48KHz source?
Maybe that´s the point.
This afternoon I will checkout a 16/48 file maybe the trouble is gone :(

[ursamtl]

Yes, for testing I used both 44.1 and 48kHz files at bit depths of both 16 and 32 bits.

You might double-check the layout as well. I have noticed that when dealing with VST plugins, Bidule sometimes retains previous settings. I've had some strange sonic delays occur with a couple of VST plugins that did not have these settings when I used them. I found that I had to use the Bidule Replace command on the VSTs, replacing them with new instances of themselves. Could it be that the mda dither plugin I used on the surround channels in that layout is somehow being loaded with some sort of a boost? I don't know, just a possibility.

Anyway, as I said, the new version is better and will be available very soon so don't spend too much time with this one.

Steve.

[daphy]

Anyway, as I said, the new version is better and will be available very soon so don't spend too much time with this one.
okay, I will wait :rolleyes: maybe you need someone for beta testing...

[oayz]

Guys,

As a newcomer, first let me thank all of you for the great job on 2->5.1. I'm looking from a bit different angle - instead of converting music, I need this to convert my home made video sound into 5.1

Anyway to start with it would be helpful to know which of 40 methods proposed and published here are considered to be the best. I not trying to establish the champion but at least to focus on most promising methods. I've heard that there was some kind of blind test done but can find neither test materials nor results.

May be you can setup some kind of voting system. It would be also interesting to know the history, roots, pros and cons of each methods

Again THANKS for putting this all together:

http://www.needfulthings.webhop.org/

[ursamtl]

Hi oayz and welcome. The problem with evaluating these methods is that the results are subjective. What sounds good to one person might sound like crap to the next! Plus as was pointed out recently, it's only natural that people will tend to prefer and promote methods they themselves have developed. If listening tests were performed, they'd have to be carried out and reported by completely independent third parties. If one of the method authors conducted listening tests and then--surprise, surprise--his or her method won by an overwhelming margin, then the credibility of the whole thing would crash! ;)

One approach that might work in your case would be to look at the process technically. Since your goal is to convert your existing movie soundtracks to 5.1, we might assume that at least some of them are in Dolby Surround. If this is the case, then one of the GraphEdit/DPLII DirectX-based methods such as JoshBM's DPLII (http://forum.doom9.org/showthread.php?s=&threadid=83384) or Bleo's original DPLII method (http://forum.doom9.org/showthread.php?s=&threadid=57736) should provide good results. The Dolby Pro Logic decoding will extract the matrixed surround information as well as the centered dialog. The results will of course depend on how well the soundtracks were encoded.

For plain stereo to 5.1 conversion, I prefer the new V.I VST (http://forum.doom9.org/showthread.php?s=&threadid=85446) I just released since it's designed for stereo material from the ground up and offers you some good adjustment options. It can even provide some good results from Dolby encoded material, since it extracts much of its surround information from the same part of the signal (left minus right) that stores the Dolby surround information. It doesn't process the information in the same way as DPL but it does provide a nice big stereo surround field.

If you have Adobe Audition, you might also try one of the methods for it listed in the Guide list (http://forum.doom9.org/showthread.php?s=&threadid=83752). Some of Audition's filtering choices for isolating dialog or removing it, might prove quite useful for processing movie soundtracks.

Probably the best thing would be to sort the movies into two batches: Dolby Surround and plain stereo. Pick a movie or two from each group and do a trial run on some representative scenes. See what the results are, adjust, experiment and find out what works best for you. Then of course, come back here and tell us about your results!

Plus if you have any questions, don't be shy! :)

Regards,
Steve.

[oayz]

ursamtl, thanks for the reply. I did try V.I., SAD51 and couple more VST based methods, as well as WinDVD DPL decoder based DirectShow method. They sounded good, though I felt like center channel separation was weak - there were still significant presence in both left and right channels. I also tried EoH voice_center method which has very good center separation but rears somehow were identical (maybe I did something wrong). Basically I was asking for help in navigating this plurality of existed methods.

I think I confused you with my setup. Source is pure stereo, not a 5.1 nor DPL. It's home made movie from DV camcorder, so sound is stereo, pretty bad stereo. The idea is to convert it to 5.1 when add music background (also mainly converted from stereo to 5.1), add comments and pin point positioning where appropriate. For the last 2 operations I use 5.1 positioning tools (Vegas Sonic).

So far for the first two tasks I use stupid method - pass stereo through my Onkyo 601 reciver and capture 6 channels. Onkyo has very good so called "5 channel stereo" mode and I'm happy with the result but the whole procedure is a mess. That's why I started to look for a software capable to do same things as Onkyo and found this and other Doom9 threads related to 2->5.1 conversition.

Anyway, I'm sure I need to spend more time listening before I do writing :-)

[ursamtl]

Oayz,

You might try the new V.I VST I just posted since it's far more advanced than the original V.I. The latter was just a work in progress I decided to post because people kept asking me for it. The new one has a lot more potential for your situation, especially with one of the Plogue Bidule layouts that add reverb to the rears. For example, one good choice would be the Bidule layout with the SIR plugin on the rears and loaded with one of the Lexicon 960 Ambience impulse responses available for free download at www.noisevault.com. Lower the limiting threshold on the center channel to perhaps -3 to -4.5dB (the amount will be displayed in the bottom corner of the 5.1 Output dialog box) to boost and compress dialog in that channel (the limiter keeps it from overloading), then move the Width Correction control on V.I to the right to give the front more spread (you could also experiment with the Front Ambience control for this). Depending on the bass content of your source, you could also lower the threshold on the LFE channel. This compresses and tightens the deep bass sounds. Combine these with the reverb on the rears, and you should end up with a really big "5.1 soundtrack-like" sound.

Steve.

[oayz]

Cool! Thanks for suggestions. I'll try and let you know how it works.

[oayz]

Ursamtl,
I guess I'm doing something wrong but the output of V.I.VST SIR is 6ch silent wav. It's not all zeroes but when I open it with SoftEncode I've got 6 flat lines. I did try Offline mode as well as 16, 32 and 24 bit outputs. Any ideas?

As a sanity check I've tried other modules (5.1SAD, VoiceCenter) and got nice 6ch waves.

It looks like some VST are not thound but I did unpack your distribution.

Thanks in adavnce for help or suggestions!

[ursamtl]

Originally posted by oayz
Ursamtl,
I guess I'm doing something wrong but the output of V.I.VST SIR is 6ch silent wav. It's not all zeroes but when I open it with SoftEncode I've got 6 flat lines. I did try Offline mode as well as 16, 32 and 24 bit outputs. Any ideas?

As a sanity check I've tried other modules (5.1SAD, VoiceCenter) and got nice 6ch waves.

It looks like some VST are not thound but I did unpack your distribution.

Thanks in adavnce for help or suggestions!

Oayz,

I replied to this message in http://forum.doom9.org/showthread.php?s=&postid=573315#post573315 because that's the thread for the VST plugin versions of V.I and I thought it would be helpful for others who might run into the same problem as yours.