Removing center stereo information....

[Raveshaper]

If I understand, you need something that can analyze a stereo pair and compare the two channels. When it finds differences between the two within a definable range, it sends those portions of the audio waveform to a "sides" output. Everything else gets sent to a "center" output. For the greatest accuracy, you would need to intelligently isolate the fundamental waveform of centered sounds and separate them from the others (importing stems and remixing, essentially). I didn't think real-time audio handling of this type was possible in the rack. Can RX4 even do this?

[selig]

If I understand, you need something that can analyze a stereo pair and compare the two channels. When it finds differences between the two within a definable range, it sends those portions of the audio waveform to a "sides" output. Everything else gets sent to a "center" output. For the greatest accuracy, you would need to intelligently isolate the fundamental waveform of centered sounds and separate them from the others (importing stems and remixing, essentially). I didn't think real-time audio handling of this type was possible in the rack. Can RX4 even do this?

What you're describing is currently "science fiction" IMO. There are far too many unanswered questions here, for example, what do you do with the sounds that are half-way between the center and one side - which "output" do you send these sounds to (assuming you could possibly isolate them)?

Besides all that, if you did have the technology to do what you describe above, you would also be able to separate all tracks to their original individual tracks. And if you can do that, then you'd probably not bother with the "center/sides" stuff!

[Raveshaper]

That's essentially what I meant. You would need to isolate sounds according to their position in stereo space, but to do that accurately you would essentially be panning and separating during the mixing phase.

The key to being able to do this with an imported audio track is to compare the waveform of the left channel with the right channel. Oscillations that match are centered, meanders within the oscillations that have more prominent amplitudes on one channel than the other belong to that side, and vice versa. To get this to work, you would have to introduce a DC offset to make stereo isolation possible, I think.

Can you introduce DC offset in audio? Maybe some Thor pass-through magic?

[normen]

What you're describing is currently "science fiction" IMO. There are far too many unanswered questions here, for example, what do you do with the sounds that are half-way between the center and one side - which "output" do you send these sounds to (assuming you could possibly isolate them)?

Besides all that, if you did have the technology to do what you describe above, you would also be able to separate all tracks to their original individual tracks. And if you can do that, then you'd probably not bother with the "center/sides" stuff!

[selig]

What you're describing is currently "science fiction" IMO. There are far too many unanswered questions here, for example, what do you do with the sounds that are half-way between the center and one side - which "output" do you send these sounds to (assuming you could possibly isolate them)?

Besides all that, if you did have the technology to do what you describe above, you would also be able to separate all tracks to their original individual tracks. And if you can do that, then you'd probably not bother with the "center/sides" stuff!

Exactly - why bother with just the center when there's tools like this (which in a few years may actually produce sounds that are useful!).

[normen]

Exactly - why bother with just the center when there's tools like this (which in a few years may actually produce sounds that are useful!).

Actually theres one like this that uses "conventional" EQing and L/R correlation, not FFT-based like this which causes the ugly glass noise. Can't remember the name though..

[ScuzzyEye]

That's essentially what I meant. You would need to isolate sounds according to their position in stereo space, but to do that accurately you would essentially be panning and separating during the mixing phase.

The key to being able to do this with an imported audio track is to compare the waveform of the left channel with the right channel. Oscillations that match are centered, meanders within the oscillations that have more prominent amplitudes on one channel than the other belong to that side, and vice versa. To get this to work, you would have to introduce a DC offset to make stereo isolation possible, I think.

Can you introduce DC offset in audio? Maybe some Thor pass-through magic?

I almost have something working (as an RE, as I think along the lines of the math involved, not how to find those math functions in Reason devices). It's not going to be perfect, but I can see it being useful for some things.

The DC offset is thinking along the right lines. I'm using full-wave rectification (absolute value), with 5 ms smoothing, to know the over-all amplitude of the left and right channels.

I'll post more details once I have a bug solved.

[ScuzzyEye]

I found the bug. I knew I had a copy and paste error somewhere, but with so many lines of l's and r's (for left and right), I just wasn't able to see the one where I had an l that should have been r (I knew that would be the case, because the right channel wasn't updating correctly). That's all taken care of now.

So, basically it is a Dolby ProLogic II decoder. I found this document http://www.pacificav.com/library/dolby% ... ration.pdf and pretty much implemented it as he describes. I've got it working pretty well. I want to add controls for how much cancellation occurs between the left/right vs. center, and the front vs. surround. I don't think it's possible to get complete cancellation, because of all the reasons covered in this thread, and additional cross-talk is introduced from left to right. So this isn't a perfect filter of any kind, but it is a nice effect, that I've wanted to have in the rack for a long time (when I listen to music on my home theater system, I always use ProLogic II decoding).