Hey folks. I've found this thing that I think is interesting, and I'm trying to understand how it works. I think DSP is fun and cool, but don't have any education in the how-it-works department.
Here's the thing:
Stereo to 5.1 Channel Converter in Max/MSP
I'm particularly interested in the LMS-based upmixing section, mostly about center channel synthesis. I wasn't able to download the MAX patch from the link, but but I was able to recreate the important bits in MAX, once I figured out the gen~ function. I emailed the author about the broken DL link, but no response.
(Max / MSP What is Max? | Cycling '''74 pretty cool tool for sound monkeying on yer computer. Free evaluation.)
From the link:
"The least-mean-square (LMS)-based method uses a decorrelation technique using the LMS algorithm. One of the original stereo channels is used as the desired signal d👎, and the other is considered as the input, x👎, of the FIR filter. The output y👎 represents the correlated part, which is used as the center channel, whereas the error e👎 that is the uncorrelated part is considered as the surround channel."
So I though that center channel y👎 would contain the common content and I could subtract that from the L and R channel (and double it) and play
(Left only content) - (L/R common content) - (Right only content)
across a 3 channel front and it would be cool.
But it doesn't work like that. Why? I made a patch with a tone generator with a L/R panner on front feeding into the modified LMS upmixer (ctr subtracted from L and R), running into a set of meters on the upmixers output, and it doesn't seem to function like I expected. (It doesn't seem to work right on listening to music either.) In fact, the REAR channels are set up as Left minus Center, hmmm.
Trying to figure out how it works leads me to more detailed questions, but I figure I'll try to get some educated individual to take the bait here before I spend a lot of column-inches laying out my ignorance. This is already getting into TL-DR territory.
Anyone out there interested in explaining this thing?
Here's the thing:
Stereo to 5.1 Channel Converter in Max/MSP
I'm particularly interested in the LMS-based upmixing section, mostly about center channel synthesis. I wasn't able to download the MAX patch from the link, but but I was able to recreate the important bits in MAX, once I figured out the gen~ function. I emailed the author about the broken DL link, but no response.
(Max / MSP What is Max? | Cycling '''74 pretty cool tool for sound monkeying on yer computer. Free evaluation.)
From the link:
"The least-mean-square (LMS)-based method uses a decorrelation technique using the LMS algorithm. One of the original stereo channels is used as the desired signal d👎, and the other is considered as the input, x👎, of the FIR filter. The output y👎 represents the correlated part, which is used as the center channel, whereas the error e👎 that is the uncorrelated part is considered as the surround channel."
So I though that center channel y👎 would contain the common content and I could subtract that from the L and R channel (and double it) and play
(Left only content) - (L/R common content) - (Right only content)
across a 3 channel front and it would be cool.
But it doesn't work like that. Why? I made a patch with a tone generator with a L/R panner on front feeding into the modified LMS upmixer (ctr subtracted from L and R), running into a set of meters on the upmixers output, and it doesn't seem to function like I expected. (It doesn't seem to work right on listening to music either.) In fact, the REAR channels are set up as Left minus Center, hmmm.
Trying to figure out how it works leads me to more detailed questions, but I figure I'll try to get some educated individual to take the bait here before I spend a lot of column-inches laying out my ignorance. This is already getting into TL-DR territory.
Anyone out there interested in explaining this thing?