neebis said:I see you used ATSurround active matrix to do the upmix. That seems to be a plugin for foobar2000 or Winamp. Did you pipe the audio through one of those or use some other method? Or is there another version of ATSurround? Can you explain your process exactly? What settings and setup did you use for the relevant applications?
Yes, I used the foobar2000 plugin. I used the default active matrix (dpl2) 5.0 settings and ran h_h's edited wavs through it and saved as a multichannel wav, then split that into mono wavs.
Did you use any of the files from the wav_to_5_1_upmixer.rar you posted earlier for the final work or other tools?
No, that was just something I found early on when trying to find ways to do upmixing, but it turned out it wasn't very good.
Is there a generational loss or potential artifacts created by the decoder process separating the Prologic channels?
Well, depends on what you consider a generation loss, there wasn't a 'lossy' step per se, but it's probably difficult to get the same bit-for-bit 2ch file back from the 5ch file..
I'm not sure if there are artifacts inherent in active matrix decoding; atsurround didn't produce any obvious artifacts to my ears. There is another upmixer called fsurround which uses libfftw3f-3.dll and it left very obvious artifacts in the channels.
Is there is a slight degradation of the audio when comparing this multi-step process to the one step of encoding the Belbecus DC48 mix directly to 2 channel AC3? Should it be noticeable or not? Would be it be like a very mild version of the artifacts caused by re-encoding an MP3 over and over a few generations?
Depending on the bitrate used when encoding the 2.0 file, the 5.1 will likely have a lower bitrate for each channel, so it's possible there are more obvious 'lossy artifacts' in comparison. I used the eac3 encoder with standard settings, I don't really know if there's another encoder considered to be better.
If you're hearing artifacts in the upmixed ac3, one way to discern if those artifacts are from the upmixing process or the ac3 encoding would be to compare it to the 1536kbps dts, which should be transparent to the wavs.