• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Red Hot Ruffled Chili Feathers

Jorj

Active Member
Forum Donor
Joined
Jan 26, 2018
Messages
293
Likes
343
Location
Washington, DC
I'm certainly going to aggravate a number of folks with this post, but in the interest of science, let's discuss this like rational audiophiles instead of knee-jerk audiophools.

I have both the original Rick Rubin 16\44.1 boosted CD to FLAC rips, as well as high quality 24\192 vinyl to FLAC rips of RHCP's Stadium Arcadium. This CD gets bagged on by EVERYONE for being compressed and brickwalled to death. I keep going back and forth between them, and you know what? The CD sounds better. Once your eyes roll back down from the top of your skull, hear me out.

The tools we use to determine Dynamic Range, are, in my humble opinion, faulty. We are painting a wide range of albums with the same broad brush as those that are terribly mastered because the utility we leverage to tell us what the DR is cannot distinguish between properly boosted levels and mashed up crap. I will offer up some evidence for my thinking.

Aside from the usual DR Meter utility, our standard method for finding if a track is compressed too much is to look at the waveform in something like Audacity. Here is the CD version of Slow Cheetah, a fave of mine.

1543521869535.png


Looks pretty bad, right? The louder sections are all mashed up against the limits. Here is the LP version...

1543521976723.png


Better, right? Lots of headroom, visibly apparent. This should sound better. But it does not sound better.

Here is what bugs me about this method of judging DR, and maybe I'm wrong about this. This waveform representation of the music shows nothing about the actual frequency spread of the audio waveforms. It just shows relative energy, unless I miss my guess. Now, sure, there is MORE energy in the CD version, but that is because the recording engineer made some decisions about how best to EQ the frequencies and boost up some of the tracks in relation to others to smooth out the sound. Some advanced algorithms are used in that process making sure the music does not actually go into clipping, but honestly, I don't think there is much of that going on. This waveform-only view of sound is super deceptive. Let's look at this another way.

Here is a VU view of Slow Cheetah from the loudest section of the song, CD version, @3:50...

1543522612069.png


Same timestamp from the LP version....

1543522769244.png


Now, according to my version of Dynamic Range Meter for Foobar2000, the LP version has a DR of 13, while the CD version has a DR of 5. This has all the armchair audiophiles up in arms, demanding that Anthony and Flea kneecap Rick Rubin and get Steve Hoffman to issue his remaster in CD format. I get that the LP rip has issues, lots of translations going on here, but come on. The CD version is visibly more even, has much nicer dynamics and no huge midbass hump. So what gives? Why is our understanding of dynamic range so flawed? Maybe it is my understanding that is flawed (and my ears) but something stinks here.

Please don't take this to mean that there are no badly mastered digital hack jobs. In fact, as I understand it, the same guy that turned Metallica's Death Magnetic into even worse-sounding hammered shit did this album also. But in this instance, I think we might want to take a deeper look and advance our understanding. Yay science! I stand ready for my beat-down.
 
OP
Jorj

Jorj

Active Member
Forum Donor
Joined
Jan 26, 2018
Messages
293
Likes
343
Location
Washington, DC
Two quick observations....first, I think Death Magnetic would have sucked regardless of how it was mastered, and it is possible, in the interest of fairness, that my LP rips are super-flawed.
 

pwjazz

Addicted to Fun and Learning
Forum Donor
Joined
Sep 21, 2018
Messages
507
Likes
746
I'm far from an expert on such topics, but I spend a fair amount of time looking at real-time spectrum analysis of my music when messing with my EQ settings, and I'm used to seeing something much closer to your LP shot, with most energy in the bass and low midrange and then decreasing into the treble. Here's a random shot from Jennifer Warnes' "First We Take Manhattan" from the audiophile favorite album "Famous Blue Raincoat". A lot of it looks like this.

Screenshot_2018-11-29-15-05-07.png
 
OP
Jorj

Jorj

Active Member
Forum Donor
Joined
Jan 26, 2018
Messages
293
Likes
343
Location
Washington, DC
I'm far from an expert on such topics, but I spend a fair amount of time looking at real-time spectrum analysis of my music when messing with my EQ settings, and I'm used to seeing something much closer to your LP shot, with most energy in the bass and low midrange and then decreasing into the treble. Here's a random shot from Jennifer Warnes' "First We Take Manhattan" from the audiophile favorite album "Famous Blue Raincoat". A lot of it looks like this.

How did you come by that file? CD rip, LP or MP3 download?
 

pwjazz

Addicted to Fun and Learning
Forum Donor
Joined
Sep 21, 2018
Messages
507
Likes
746
Also, and this may mean nothing, your LP spectrum looks more like pink noise and your CD spectrum like white noise.
 
OP
Jorj

Jorj

Active Member
Forum Donor
Joined
Jan 26, 2018
Messages
293
Likes
343
Location
Washington, DC

pwjazz

Addicted to Fun and Learning
Forum Donor
Joined
Sep 21, 2018
Messages
507
Likes
746
For some more perspective, here's some random noise rock that I downloaded from the Free Music Archive:

1543526455560.png


And here's the spectrum:

1543526479648.png


One could say that looks nicely balanced and full. But, it is in practice not that far removed from white noise.

EDIT - If I analyze the spectrum for a quieter section towards the end, it shows that nicer pink-noise like downward slope that I associate with "pleasant".

1543526597050.png
 
OP
Jorj

Jorj

Active Member
Forum Donor
Joined
Jan 26, 2018
Messages
293
Likes
343
Location
Washington, DC
Also, and this may mean nothing, your LP spectrum looks more like pink noise and your CD spectrum like white noise.

That is what I'm saying. The tools we use to determine whether or not music has been over-compressed don't seem to lend themselves to visual analysis. DR seems meaningless when visually, music is difficult to distinguish from noise. The spectrum graph is a snapshot of a single point in time, and the waveform from Audacity just tells you when the music is loud and when it is quiet, not much else.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Most masters of rock and pop have a more pink-noise like spectrum, with a steeper drop-off above 4 to 5 KHz. This is considered standard/desirable by mastering engineers IME, and tends to follow from the spectra for things like human voices and guitars.

I'm surprised to see such a huge difference in spectra between the spectra for the vinyl and CD versions of Stadium Arcadium at that particular timestamp. Are you sure that is the same moment in the music being represented? Would seem very unlikely.
 
OP
Jorj

Jorj

Active Member
Forum Donor
Joined
Jan 26, 2018
Messages
293
Likes
343
Location
Washington, DC
Most masters of rock and pop have a more pink-noise like spectrum, with a steeper drop-off above 4 to 5 KHz. This is considered standard/desirable by mastering engineers IME, and tends to follow from the spectra for things like human voices and guitars.

I'm surprised to see such a huge difference in spectra between the spectra for the vinyl and CD versions of Stadium Arcadium at that particular timestamp. Are you sure that is the same moment in the music being represented? Would seem very unlikely.

I got them as close as I could, but spectra are hard to time-align exactly with the tools I have at my disposal. In general, yes, the LP version has a distinct bass and midbass hump in comparison to the CD version. This is true for all songs on that album. DESPITE that big injection of energy, DR Meter thinks it has more dynamic range, because it is looking at the overall waveform, not the actual audio spectra.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
I got them as close as I could, but spectra are hard to time-align exactly with the tools I have at my disposal. In general, yes, the LP version has a distinct bass and midbass hump in comparison to the CD version. This is true for all songs on that album. DESPITE that big injection of energy, DR Meter thinks it has more dynamic range, because it is looking at the overall waveform, not the actual audio spectra.

Fair enough. But to me the far more likely explanation is that the spectrum from your CD rip is taken in between hits on the kick drum or bass notes, while the LP spectrum falls right on a kick and/or bass note.

I think you really need to average the spectrum across a second or more at the same moment in the recording to make anything meaningful of it.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Further supporting my theory is the fact that the peaks roughly align.

If the averaged spectra really were that different, the CD master would sound horribly tinny and thin.

No properly mastered rock/pop music has an average spectrum like that during a passage where the kick and bass guitar are playing.
 
OP
Jorj

Jorj

Active Member
Forum Donor
Joined
Jan 26, 2018
Messages
293
Likes
343
Location
Washington, DC
I think you really need to average the spectrum across a second or more at the same moment in the recording to make anything meaningful of it.

I am not in a position to average a spectrum over time, but just from visually watching the VU, the LP version really does have the appearance of higher levels at lower frequencies. Further, I can attest without reservation that the CD rip does not sound thin and tinny. It has loads of bass, but does not look as 'smoothed' as the LP version.
 

pwjazz

Addicted to Fun and Learning
Forum Donor
Joined
Sep 21, 2018
Messages
507
Likes
746
I got them as close as I could, but spectra are hard to time-align exactly with the tools I have at my disposal

If you use Audacity, you can select the entire track and generate a composite spectrum without having to worry about time aligning.

1543528585905.png
 
OP
Jorj

Jorj

Active Member
Forum Donor
Joined
Jan 26, 2018
Messages
293
Likes
343
Location
Washington, DC
Just to redirect a bit, my contention is that our usage of Dynamic Range tools like those found in Foobar and MAAT are flawed and ultimately reveal next to nothing when they are faced with a well-mastered digital recording that used some compression. The numbers they provide seem to be not just worthless, but deeply misleading.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
I am not in a position to average a spectrum over time, but just from visually watching the VU, the LP version really does have the appearance of higher levels at lower frequencies. Further, I can attest without reservation that the CD rip does not sound thin and tinny. It has loads of bass, but does not look as 'smoothed' as the LP version.

I'd be interested in looking at the rips, or say the same 30 seconds of each, if you can be bothered uploading them? :)

And I'm not suggesting the CD version should sound thin and tinny, just that the screenshot you've taken may be misleading.

EDIT: the main difference between the two spectra you've posted seems to be that the CD is mixed with less sub bass (I'm looking at the peak levels). You can't glean much about the dynamic range from those two snapshots in themselves.
 
Last edited:
OP
Jorj

Jorj

Active Member
Forum Donor
Joined
Jan 26, 2018
Messages
293
Likes
343
Location
Washington, DC
Top Bottom