(With apologies to Steve Hoffman for paraphrasing his thread title.)
I've been in the process of ripping my surround collection to multichannel FLAC, and spurred on by the success of fixing the swapped front/rear channels on Neil Young's 'Harvest' I resolved to look at some of the other discs in my collection with wonky mixes that have troubled me. I've wanted to do this for a long time, but the watermarking on DVD-A always prevented me from being able to play my corrected mixes back - now with FLAC this is no longer an issue.
The disc at the top of my list was George Benson's 'Breezin', which I got on release day some 10 years or so ago. It troubled me from the first listen... seemingly drenched in reverb, with very little bottom end to speak of - I remember being really annoyed at the fact that the bass drum 'popped' instead of thumped. The whole thing was baffling to me though, given that the album was remixed for 5.1 by the original producer. Not just that they could end up with a bad mix, but that they end up with a 5.1 mix that sounded so unlike the original stereo mix, which has great bass response and isn't drenched in reverb.
So I ripped the album to 5.1 WAV files using Foobar2000 and loaded the first track in to Nuendo and this is what I saw:
It was immediately obvious just by looking at the waveforms that the centre and LFE channels had been turned way down, but that there was content there.
So I put my headphones and had a listen to those two tracks, and I was shocked with what I found: the centre channel has DRY lead guitar, vocals, bass guitar, and drums in it, and the LFE channel (less shockingly) has bass guitar and drums in it. I started to think that maybe the same fate had befallen this disc that befell the Led Zeppelin 'How The West Was Won' disc, that after the 5.1 mix was done, for whatever reason the mastering engineer decided to second guess the mixing engineer and drastically altered the level of the centre channel.
I wanted to balance the centre and LFE channels in relation to the other speakers in a (to some degree) scientific method, rather than just subjectively guessing at what the levels should be. To do this I used a wonderful free VST plugin that Neil Wilkes turned me on to called Voxengo SPAN. It does realtime visual audio spectral analysis, and it allows you to overlay one audio signal on top of another, so you can compare them relative to each other.
So I made two audio groups, one with the Centre & LFE channels (yellow curve), and another with the Front Left and Front Right channels (green curve), and piped them through SPAN. This is what I saw:
It's clear from the shape of the curves that there's a ton of common content between these channels, but just that the C+LFE channels were way lower in volume. So I first tried boosting the volume of the C+LFE channels so that the volume of the lead guitar was the same in the centre channel as it was in the FL & FR channels. So I did this and had a listen to the mix and it didn't sound right - the front soundstage didn't have any depth, it was just a wall in your face, and the bass was overpowering. So I went back to the drawing board and decided instead to match the bass levels in the C+LFE with the FL+FR channels. Once I did this, my SPAN readout looked like this:
As you can see the bass matches up exactly, and the volume of the rest of the frequency spectrum in the C+LFE is a few dB lower than the FL+FR channels. I had a listen to the mix like this and OH MY GOD it's a revelation! It sounds like the proper 5.1 version of the stereo album we should've got, rather than the abomination we did get. The reverby echoiness is gone, and the bottom end is full and solid!
Strangely, different tracks required slightly different boosts, but all were within the 12-14dB range, with a couple of notable exceptions:
Track 6 ('Lady') has the orchestral intro mixed at the correct volume in all 6 channels. I had to make a split between the orchestral intro and the rest of the track on the centre and LFE channels because they still needed their 12-14dB boosts even though the orchestral intro was correct.
Track 6 - note the identical waveform in all 6 channels.
Track 8 ('Down Here On The Ground') has an almost correct LFE channel, but not quite. Look at how much more energy there is in the LFE track in comparison to the previous two tracks.
Track 8 - note the much louder LFE track...none of the other songs on the disc are like this.
For anyone who wants to attempt this on their own, here's a list of the channel alterations I made (in dB):
TRACK___C____LFE
-----------------------
1_______13___13
2_______13.5_13.5
3_______14___14
4_______14___14
5_______13___13
6_______13___13 (except for string intro)
7_______14___10.5
8_______12___3.5
Spectral analysis revealed some interesting results too - there's pretty much no audio content about 18kHz, aside from some bands of ultrasonic noise in the 30kHz to 40kHz range, which accounts for the somewhat dull sound of this album. I have CD's that have frequency extension higher than 18kHz!
And last but not least, here are the DR numbers from Foobar's DR meter, which are very good:
--------------------------------------------------------------------------------
Analyzed: George Benson / Breezin' (96/24 5.1 Surround)
--------------------------------------------------------------------------------
DR Peak RMS Duration Track
--------------------------------------------------------------------------------
DR15 -2.45 dB -20.52 dB 5:42 01-Breezin'
DR15 -2.43 dB -22.25 dB 8:05 02-This Masquerade
DR15 -2.57 dB -21.86 dB 5:11 03-Six To Four
DR14 -2.42 dB -20.32 dB 6:51 04-Affirmation
DR14 -1.91 dB -20.47 dB 7:00 05-So This Is Love?
DR14 -1.71 dB -21.18 dB 6:02 06-Lady
DR15 -0.74 dB -21.85 dB 6:11 07-Shark Bite
DR15 -0.38 dB -20.64 dB 9:05 08-Down Here On The Ground
--------------------------------------------------------------------------------
Number of tracks: 8
Official DR value: DR14
Samplerate: 96000 Hz
Channels: 6
Bits per sample: 24
Bitrate: 6823 kbps
Codec: FLAC
================================================================================
All in all, pretty interesting stuff. Anyone else brave enough to attempt the channel alterations I made? I'd be interested in what other people think.
I've been in the process of ripping my surround collection to multichannel FLAC, and spurred on by the success of fixing the swapped front/rear channels on Neil Young's 'Harvest' I resolved to look at some of the other discs in my collection with wonky mixes that have troubled me. I've wanted to do this for a long time, but the watermarking on DVD-A always prevented me from being able to play my corrected mixes back - now with FLAC this is no longer an issue.
The disc at the top of my list was George Benson's 'Breezin', which I got on release day some 10 years or so ago. It troubled me from the first listen... seemingly drenched in reverb, with very little bottom end to speak of - I remember being really annoyed at the fact that the bass drum 'popped' instead of thumped. The whole thing was baffling to me though, given that the album was remixed for 5.1 by the original producer. Not just that they could end up with a bad mix, but that they end up with a 5.1 mix that sounded so unlike the original stereo mix, which has great bass response and isn't drenched in reverb.
So I ripped the album to 5.1 WAV files using Foobar2000 and loaded the first track in to Nuendo and this is what I saw:

It was immediately obvious just by looking at the waveforms that the centre and LFE channels had been turned way down, but that there was content there.
So I put my headphones and had a listen to those two tracks, and I was shocked with what I found: the centre channel has DRY lead guitar, vocals, bass guitar, and drums in it, and the LFE channel (less shockingly) has bass guitar and drums in it. I started to think that maybe the same fate had befallen this disc that befell the Led Zeppelin 'How The West Was Won' disc, that after the 5.1 mix was done, for whatever reason the mastering engineer decided to second guess the mixing engineer and drastically altered the level of the centre channel.
I wanted to balance the centre and LFE channels in relation to the other speakers in a (to some degree) scientific method, rather than just subjectively guessing at what the levels should be. To do this I used a wonderful free VST plugin that Neil Wilkes turned me on to called Voxengo SPAN. It does realtime visual audio spectral analysis, and it allows you to overlay one audio signal on top of another, so you can compare them relative to each other.
So I made two audio groups, one with the Centre & LFE channels (yellow curve), and another with the Front Left and Front Right channels (green curve), and piped them through SPAN. This is what I saw:

It's clear from the shape of the curves that there's a ton of common content between these channels, but just that the C+LFE channels were way lower in volume. So I first tried boosting the volume of the C+LFE channels so that the volume of the lead guitar was the same in the centre channel as it was in the FL & FR channels. So I did this and had a listen to the mix and it didn't sound right - the front soundstage didn't have any depth, it was just a wall in your face, and the bass was overpowering. So I went back to the drawing board and decided instead to match the bass levels in the C+LFE with the FL+FR channels. Once I did this, my SPAN readout looked like this:

As you can see the bass matches up exactly, and the volume of the rest of the frequency spectrum in the C+LFE is a few dB lower than the FL+FR channels. I had a listen to the mix like this and OH MY GOD it's a revelation! It sounds like the proper 5.1 version of the stereo album we should've got, rather than the abomination we did get. The reverby echoiness is gone, and the bottom end is full and solid!
Strangely, different tracks required slightly different boosts, but all were within the 12-14dB range, with a couple of notable exceptions:
Track 6 ('Lady') has the orchestral intro mixed at the correct volume in all 6 channels. I had to make a split between the orchestral intro and the rest of the track on the centre and LFE channels because they still needed their 12-14dB boosts even though the orchestral intro was correct.

Track 6 - note the identical waveform in all 6 channels.
Track 8 ('Down Here On The Ground') has an almost correct LFE channel, but not quite. Look at how much more energy there is in the LFE track in comparison to the previous two tracks.

Track 8 - note the much louder LFE track...none of the other songs on the disc are like this.
For anyone who wants to attempt this on their own, here's a list of the channel alterations I made (in dB):
TRACK___C____LFE
-----------------------
1_______13___13
2_______13.5_13.5
3_______14___14
4_______14___14
5_______13___13
6_______13___13 (except for string intro)
7_______14___10.5
8_______12___3.5
Spectral analysis revealed some interesting results too - there's pretty much no audio content about 18kHz, aside from some bands of ultrasonic noise in the 30kHz to 40kHz range, which accounts for the somewhat dull sound of this album. I have CD's that have frequency extension higher than 18kHz!

And last but not least, here are the DR numbers from Foobar's DR meter, which are very good:
--------------------------------------------------------------------------------
Analyzed: George Benson / Breezin' (96/24 5.1 Surround)
--------------------------------------------------------------------------------
DR Peak RMS Duration Track
--------------------------------------------------------------------------------
DR15 -2.45 dB -20.52 dB 5:42 01-Breezin'
DR15 -2.43 dB -22.25 dB 8:05 02-This Masquerade
DR15 -2.57 dB -21.86 dB 5:11 03-Six To Four
DR14 -2.42 dB -20.32 dB 6:51 04-Affirmation
DR14 -1.91 dB -20.47 dB 7:00 05-So This Is Love?
DR14 -1.71 dB -21.18 dB 6:02 06-Lady
DR15 -0.74 dB -21.85 dB 6:11 07-Shark Bite
DR15 -0.38 dB -20.64 dB 9:05 08-Down Here On The Ground
--------------------------------------------------------------------------------
Number of tracks: 8
Official DR value: DR14
Samplerate: 96000 Hz
Channels: 6
Bits per sample: 24
Bitrate: 6823 kbps
Codec: FLAC
================================================================================
All in all, pretty interesting stuff. Anyone else brave enough to attempt the channel alterations I made? I'd be interested in what other people think.