Disclaimer: I ripped this SACD using a modified ps3 and then converted the DSD files to 88.kHz/24 bit PCM using Foobar2000's SACD decoder. It's possible (or at least arguable) that this has some effect on quality, so please take these observations with a small grain of salt.
The recent discussion on the Isley Brothers 3+3 review thread caused me to pull this one out and revisit it. The rear channels on this disc are most definitely swapped. For example, on the first track 'Who's That Lady', the lead guitar comes out of the front left and rear right speakers, which is incorrect - it should be from the front left and rear left, which switching the rear speakers corrects. A diagonally panned instrument would give a SQ encoder an epileptic fit, there's no way they'd do that.
The fact that the rears were swapped caused me to wonder if there were other mastering errors with the album, ie it was a rush job and they didn't bother to do any quality control, or if they legitimately thought it should sound like that. After looking at all the tracks, I can't say I'm any more sure of the answer than when I started. Every single track has left/right channel imbalances in both the front and rear pairs, and the imbalances vary from track to track. Now I'm not sure if they were trying to alter the instrument balance a little bit to make it more pleasing (to their ears) for 5.1, or if they tried to do the balance by ear without the benefit of any kind of scope, or what.
Note that I deleted the derived centre channel because it destroys the stereo image in the front, but I kept the LFE track because I wasn't sure if what was in the LFE track was shaved off the other 4 channels and I didn't want to lose any bass response.
Let's look at the first track, 'Who's That Lady':
What becomes obvious from first look (aside from the fact that the dynamic range is obviously compressed by the shape of the waveforms) is that there's way more energy in the front speakers than the rears, which is (in my opinion) wrong. Quad mixes, especially the Columbia/CBS ones, almost always had equal power from all 4 speakers. It appears that by adding a derived center channel and reducing the power of the rears that they were trying to make the mix sound more like a modern 5.1 mix.
The next thing I noticed is the problems with the left/right balance. Here's a look at the front left (green) and front right (yellow) channels of the second track, 'Don't Let Me Be Lonely Tonight':
I think you can quite clearly see that the right channel is uniformly lower than the left one, and this happens to differing degrees on the front and back pairs on most of the tracks. When you fix these imbalances, you really notice it in the vocals, which have a pinpoint center focus and much more pleasing phantom image. Getting the front/rear balances was a little more difficult, but I used a combination of looking at waveforms using the SPAN spectral analyser plugin, and also listening to isolated front and rear pairs (FL & RL / FR & RR) on headphones until I was satisified the balance was correct. The front/rear balance may not be 100% scientifically accurate, but I don't think I'm more than half a dB out on any given track.
Also, it's worth mentioning that track 6 ('What It Comes Down To') is missing the audio in the rear channels for the first 15 seconds of the song, as you can see in the bottom two waveforms in this picture:
These are my reccomendations for channel level alterations - this is after you've swapped the rear channels so the channel assignments are correct:
TRACK|FL|FR|SL|SR
1|0dB|-1dB|+3dB|+1dB
2|0dB|+1dB|+3dB|+2.5dB
3|0dB|+0.5dB|+3dB|+3dB
4|0dB|+1dB|+2dB|+2dB
5|0dB|+1dB|+2dB|+2dB
6|0dB|-0.25dB|-1dB|-1dB
7|0dB|+1.25dB|+2dB|+2dB
8|0dB|+1.5dB|0dB|0dB
9|0dB|+2.75dB|+2dB|+2dB
The mix is mastered really hot so you should turn down your final output by 3-4dB to avoid any clipping.
Spectral analysis reveals some interesting (if disappointing) results:
Frequency response drops of abruptly at 22k, which means it's very possible this was sourced from a 44.1kHz master. This would jibe with what Tad said about Sony archiving and delivering their masters on ADAT in the 90's when HDS was releasing DTS CD's. You can also see how the dynamic range compression applied to the audio has totally flattened out the frequency response, and this is borne out by the dynamic range totals as reported by the DR plugin for foobar2000:
DR Peak RMS Duration Track
--------------------------------------------------------------------------------
DR11 -2.01 dB -15.29 dB 5:30 ?-track1
DR12 -1.00 dB -16.41 dB 4:19 ?-track2
DR12 -1.06 dB -16.21 dB 3:19 ?-track3
DR12 -0.98 dB -16.42 dB 3:05 ?-track4
DR11 -1.05 dB -14.65 dB 4:05 ?-track5
DR11 -0.75 dB -14.56 dB 4:01 ?-track6
DR12 -1.77 dB -16.57 dB 4:27 ?-track7
DR11 -0.52 dB -15.20 dB 6:10 ?-track8
DR13 -0.96 dB -18.48 dB 4:15 ?-track9
--------------------------------------------------------------------------------
Number of tracks: 9
Official DR value: DR11
Samplerate: 88200 Hz
Channels: 5
Bits per sample: 24
Bitrate: 5514 kbps
Codec: FLAC
========================
Hope you've enjoyed the latest instalment in this series. Don't forget to tip your waitress!
The recent discussion on the Isley Brothers 3+3 review thread caused me to pull this one out and revisit it. The rear channels on this disc are most definitely swapped. For example, on the first track 'Who's That Lady', the lead guitar comes out of the front left and rear right speakers, which is incorrect - it should be from the front left and rear left, which switching the rear speakers corrects. A diagonally panned instrument would give a SQ encoder an epileptic fit, there's no way they'd do that.
The fact that the rears were swapped caused me to wonder if there were other mastering errors with the album, ie it was a rush job and they didn't bother to do any quality control, or if they legitimately thought it should sound like that. After looking at all the tracks, I can't say I'm any more sure of the answer than when I started. Every single track has left/right channel imbalances in both the front and rear pairs, and the imbalances vary from track to track. Now I'm not sure if they were trying to alter the instrument balance a little bit to make it more pleasing (to their ears) for 5.1, or if they tried to do the balance by ear without the benefit of any kind of scope, or what.
Note that I deleted the derived centre channel because it destroys the stereo image in the front, but I kept the LFE track because I wasn't sure if what was in the LFE track was shaved off the other 4 channels and I didn't want to lose any bass response.
Let's look at the first track, 'Who's That Lady':
What becomes obvious from first look (aside from the fact that the dynamic range is obviously compressed by the shape of the waveforms) is that there's way more energy in the front speakers than the rears, which is (in my opinion) wrong. Quad mixes, especially the Columbia/CBS ones, almost always had equal power from all 4 speakers. It appears that by adding a derived center channel and reducing the power of the rears that they were trying to make the mix sound more like a modern 5.1 mix.
The next thing I noticed is the problems with the left/right balance. Here's a look at the front left (green) and front right (yellow) channels of the second track, 'Don't Let Me Be Lonely Tonight':
I think you can quite clearly see that the right channel is uniformly lower than the left one, and this happens to differing degrees on the front and back pairs on most of the tracks. When you fix these imbalances, you really notice it in the vocals, which have a pinpoint center focus and much more pleasing phantom image. Getting the front/rear balances was a little more difficult, but I used a combination of looking at waveforms using the SPAN spectral analyser plugin, and also listening to isolated front and rear pairs (FL & RL / FR & RR) on headphones until I was satisified the balance was correct. The front/rear balance may not be 100% scientifically accurate, but I don't think I'm more than half a dB out on any given track.
Also, it's worth mentioning that track 6 ('What It Comes Down To') is missing the audio in the rear channels for the first 15 seconds of the song, as you can see in the bottom two waveforms in this picture:
These are my reccomendations for channel level alterations - this is after you've swapped the rear channels so the channel assignments are correct:
1|0dB|-1dB|+3dB|+1dB
2|0dB|+1dB|+3dB|+2.5dB
3|0dB|+0.5dB|+3dB|+3dB
4|0dB|+1dB|+2dB|+2dB
5|0dB|+1dB|+2dB|+2dB
6|0dB|-0.25dB|-1dB|-1dB
7|0dB|+1.25dB|+2dB|+2dB
8|0dB|+1.5dB|0dB|0dB
9|0dB|+2.75dB|+2dB|+2dB
The mix is mastered really hot so you should turn down your final output by 3-4dB to avoid any clipping.
Spectral analysis reveals some interesting (if disappointing) results:
Frequency response drops of abruptly at 22k, which means it's very possible this was sourced from a 44.1kHz master. This would jibe with what Tad said about Sony archiving and delivering their masters on ADAT in the 90's when HDS was releasing DTS CD's. You can also see how the dynamic range compression applied to the audio has totally flattened out the frequency response, and this is borne out by the dynamic range totals as reported by the DR plugin for foobar2000:
DR Peak RMS Duration Track
--------------------------------------------------------------------------------
DR11 -2.01 dB -15.29 dB 5:30 ?-track1
DR12 -1.00 dB -16.41 dB 4:19 ?-track2
DR12 -1.06 dB -16.21 dB 3:19 ?-track3
DR12 -0.98 dB -16.42 dB 3:05 ?-track4
DR11 -1.05 dB -14.65 dB 4:05 ?-track5
DR11 -0.75 dB -14.56 dB 4:01 ?-track6
DR12 -1.77 dB -16.57 dB 4:27 ?-track7
DR11 -0.52 dB -15.20 dB 6:10 ?-track8
DR13 -0.96 dB -18.48 dB 4:15 ?-track9
--------------------------------------------------------------------------------
Number of tracks: 9
Official DR value: DR11
Samplerate: 88200 Hz
Channels: 5
Bits per sample: 24
Bitrate: 5514 kbps
Codec: FLAC
========================
Hope you've enjoyed the latest instalment in this series. Don't forget to tip your waitress!