Converting stereo tracks to 5.1?

[depressed]

I wonder what I would need to get it done. Just for fun, in the past I have "converted" mono to stereo files by simply copying the mono track and pasting it as another channel and then randomly apply ultra short delay to one of the "2" channels. Repeated the procedure about 50-60 times during the song randomly to the right or left channel. It's fun, try it!
Back to topic: I heard that there is a program (or better yet an algorithm) that does it by itself, automated! I really don't know how good the result of such thing might be. I just can'timagine such detailed and complex algorithm that can "recognize" various sounds/patterns and apply the encoding accordingly.
My question is: How can one break down the "original" stereo track into instruments/voices and then "move them around"???
This has been on my mind for a while. BTW, is there such software available to us mortals too? Recently I found a WAV file on the net that was DTS encoded ( it was Christina Aguilera's song!?!) but the guys that did it didn't mention what software they used and it didn't sound good, I had the feel they had just moved the whispers to the surround speakers. Anyway, am I the only one itrigued by this?

[Sir Terrence the Terrible]

I wonder what I would need to get it done. Just for fun, in the past I have "converted" mono to stereo files by simply copying the mono track and pasting it as another channel and then randomly apply ultra short delay to one of the "2" channels. Repeated the procedure about 50-60 times during the song randomly to the right or left channel. It's fun, try it!

Acturally, you didn't create a stereo track at all. You did what THX does to their rear channels to give the impression of more spaciousness. We call the psuedo stereo in the industry. They call it dynamic decorrelation. You have to have 2 channels with totally dissimular information on them to be deemed sterero.


Back to topic: I heard that there is a program (or better yet an algorithm) that does it by itself, automated! I really don't know how good the result of such thing might be. I just can'timagine such detailed and complex algorithm that can "recognize" various sounds/patterns and apply the encoding accordingly.
My question is: How can one break down the "original" stereo track into instruments/voices and then "move them around"???
This has been on my mind for a while. BTW, is there such software available to us mortals too? Recently I found a WAV file on the net that was DTS encoded ( it was Christina Aguilera's song!?!) but the guys that did it didn't mention what software they used and it didn't sound good, I had the feel they had just moved the whispers to the surround speakers. Anyway, am I the only one itrigued by this?

Since this is an area of expertise for me, I can tell you that this is not exactly a software driven process.

First you need access to the individual channels of the mix(i.e all 16 to 48 depending on how many channels in the mix). Then you need a mixer with a pan pots so you can direct the individual stems to the desired position(software or hardware related). Then you need a eq, compressors, limiters, and various other processing gear to message the tracks to taste.

You need all of this gear because once tracks have been downmixed to two channel from the various original stems, there is no effective(or cheap) way to unseperate them without the use of the original stems. Once everything has been downmixed, mixed, and mastered, at that point Dts or any other 5.1 channel codec can be used for encoding.

[depressed]

I found some info here:
http://forum.doom9.org/showthread.php?threadid=60137

[Sir Terrence the Terrible]

I found some info here:
http://forum.doom9.org/showthread.php?threadid=60137

Good find depressed. I found it rather interesting so I found and downloaded all of the necessary software, and gave it a try. I want to first preface my results with a couple of comments. Ambisonics B format is a matrix driven format. In other words it not TRUE discreet 5.1, it a matrix encoded 5.0 It works simularly to the Dolby surround encoder, but utilizes two channels instead of four to get the surround effect. What this process does is indentify out of phase signals within the stereo tracks, seperate that information from each channel, and sends it to the corresponding rear speaker. In other words the right channels out of phase signals are sent to the right rear channel, and so forth for the left front to left rear. The center channels information come from the in phase signals in the L/R mains, filtered out of those channels, and encoded to the center speaker. Much like Dolby surround. None of this is discreet however, its all matrix, with SOME of the matrixes shortcomings.

My encoding and listening.

The process of encoding is pretty simple and straightforward. I followed the instructions to the letter, and had no problems at all. I encoded the results into Dts at 48khz( I had to do everything else in 44.1khz) at a bit rate of 1509kbps. I used a discrete stereo mixdown version of a 5.1 recording I did a couple of months ago as the source to encode, and the discrete 5.1 version as the reference check for quality. Just to make things equal, I did not encode a LFE channel.

The results sounded pretty damn good. Certainly better than a Dolby surround encoding would. However, it could not match the true discrete 5.1 version it was derived from. The front channels sounded closed in, with the soundstage sounding a little shunted back to front, and definately narrowed left to right. As a result, some of the instruments/musicians sounded like they were sitting in each others lap, as opposed to next to them. The surrounds were not discrete at all, so it was no comparison to the truely discrete surrounds from the original referenced source. Although for comparison sake I folded the LFE into my L/R mains(they are more than capable of handling this task), the bass was still deeper on the original than on the encoded mix.

Overall this process sounds pretty good. It however is not 5.1 in the true since of discrete 5.1. It is a matrixed derivative of 5.0 with a optional LFE channel that is really filtered bass from the other channels. If you already have a stereo source you would like to encoded into the process, it certainly sounds better than plain stereo. But when you compare it to true 5.1 tracks, there seems to be quite of bit of loss in the transfer

One good thing from the process is it does not limit the bandwidth in the rear channels as Dolby stereo does. The flip side is that would be a problem for owners of small surrounds depending on the amount of bass in the source.

Thanks for the heads up, great program for what it does.

[depressed]

Thank you for your insights, Sir TtT. I have been playing with the idea of converting channel tracks into 5.1 for a while and I was unaware that it can't be done with amateur equipment/software. Working with ambience was fun but it's quite limited. I even tried applying EQ to the tracks to kill "the voices" in the surround speakers. That was one of the main reasons why I tried it, I was loooking for something more then stereo and was unhappy with dolby prologic variations. My conclusion is that ambisonic is more useful if it's done with tracks like techno/dance songs as the voices are (almost) non-existent in that kind of music.

[depressed]

Another find, expensive ($800) however more features are available:
http://www.steinberg.net/ProductPage_sb.asp?Product_ID=2014&Langue_ID=7

(must open in IE with allow cookies)