So a couple of years ago I said this:
More recently I've been ripping my entire collection to FLAC including DTS DVD-Vs and it was always my intention to rip this and discard the center channel. So while I was doing that I had a look at the whole thing to check for phase errors, channel assignment problems etc.
What I found (in a nutshell, more detail below) is the following:
1. The center channel was created by duplicating the front left channel (only), applying an 80ms delay and then adding an EQ boost of a few dB to it from about 100Hz up to 2500Hz. The result is that the center channel sounds a bit like a loudspeaker at a shopping mall or something, and the delay between it and the FL channel causes all kinds of weird phase issues - the phase meter i was using was fluttering back and forth like it was having a seizure or something. Normally fake center channels are created by summing the front left and right channels and then reducing the volume of the resultant derived channel, so the way they've done it here is odd to say the least. Also, with the EQ boost on the center channel it effectively makes it the loudest/most dominant channel in front of you and collapses the stereo image in to the middle - this album sounds so much better with it gone.
2. The LFE channel is out of phase (!) and way way too loud. My supposition is that because the LFE was out of phase, the more they turned it up, the more it was cancelling out the bass in the 4 main channels. So as a result they just kept cranking it up until it reached some kind of point of equilibrium, because when I corrected the phase of the LFE I had to turn it down by 20 (TWENTY) dB. There's actually nothing unique in the LFE channel (ie they didn't shave any bass off the main channels) so if you want to delete the LFE track entirely you can. I chose to keep it because even with large speakers, I think it's nice to have a little something coming out of the subwoofer if you have your speakers set to large. The 20dB reduction was just done by ear and looking at waveforms (and based on the LFE levels in other .1 tracks I've seen) so it may not be perfect, but it's in the ballpark.
3. The track index points are wrong for every single track (and by different amounts for each track) so the net result is that the first 2 or 3 notes of each song is cut off, instead appearing at the tail end of the track before. I imported all the songs in to my workstation as one long track and re-exported them so the tracks were split correctly. If you want to do this too, the track timings are as such:
Track 1: 00:00.000
Track 2: 04:06.350
Track 3: 08:32.000
Track 4: 13:03.000
Track 5: 18:10.600
Track 6: 21:30.000
Track 7: 24:56.350
Track 8: 29:23.000
Track 9: 33:37.000
Everything else checks out though, the channel assignments are correct, everything (else) is in phase, and it doesn't look like they compressed it much if at all - the DR readings are between DR11 and DR13 which puts it close to even the unremastered 80's CD issue. I'm not overly in love with the EQ - it's a little midrangey and the treble falls off steeply around 15kHz but it's probably as good as it gets for standard DTS. I'm still experimenting with EQ for this album but i think if you take a little bit out between 1kHz and 3kHz it might help a bit.
Here are a couple of screenshots that illustrate the problems I was explaining earlier:
Fig 1. showing that the center channel waveform is identical to the front left
(the red line shows where the identical time-offset peaks start)
View attachment 25818
Fig 2. showing the weird midrange boost on the center channel
(The green waveform is the front left channel, and the pink bits that stick up between 100Hz and 2500Hz are the center channel, showing where it's boosted relative to the front left.)
View attachment 25817
Fig 3. showing how LOUD the LFE track is relative to the other channels
View attachment 25819
I didn't do a screenshot of my phase meter showing that the LFE was out of phase, but you can take my word for it that it was. The phase meter goes from -1 (out of phase) to 0 (no phase correlation at all) to +1 (in phase) and the LFE track was consistently at -1 with no movement at all.
One last note, I don't think Glenn Hughes did the quad to 5.1 mix, he supervised the remixing of the bonus tracks on the stereo CD. I believe Peter Mew did the quad to 5.1 transfer which would probably explain a lot of the problems.
