DT12 CD vs HDTracks - A visual representation

[Zook]

CRAP!

While walking to work, I just realized there is a bug in my script. Specifically, it makes the comparisons between the albums somewhat invalid. I am retracting the graph for now, and will regenerate it when I get back from work. Stay tuned!
(for the technically minded: When I'm generating the histogram for each album, I'm first looking at the maximum energy, so I can rescale it so it fits into the 1,000 buckets. However, for the graphs I'm not "unscaling" it again. So, albums might be squished or expanded due to the bug)

How long will we have to wait for the PS3 patch?

[rumborak]

Fixed the bug, new graph!
Not really much has changed though, the order of albums remained the same, and DT12's quietest parts are still louder than IAW loudest.

EDIT: Interesting tidbit, the loudest section in all those albums is TDEN, 7.7 seconds in

[Calvin6s]

Can you repeat what the x and y axis represents again?

I'm thinking it is:
X = the bit representation of the db level at that frame (point in time)
Y = the # of frames that db level occurs through out the total album.

Every time I see an axis going up to about 20,000, I immediately think Hz-pitch, but I don't think that was what you were talking about.

[rumborak]

The x-axis is "RMS" value, which is essentially a measure proportional to the dB value (and thus loudness) of a section. The sections are 100ms long each.
And yeah, the y axis is how many segments had that RMS value.

[Calvin6s]

So Awake and I&W look similar with Awake just having the headroom increased, so the entire graph shifts.

Met 2 seems to have a low volume peak.  I'm guessing this is because the album songwriting has more lows, especially the spoken word parts.

Falling Into Infinity does seem to have the widest dynamic range spread.

The two Roadrunner examples seem to be really close to one another.  Lots of loud noises in a short dynamic range ... range.

[BlobVanDam]

EDIT: Interesting tidbit, the loudest section in all those albums is TDEN, 7.7 seconds in

I absolutely would have guessed that. With the cranked guitar/bass/drums in unison, you can hear it pumping the compression.

And where's SDOIT?

[rumborak]

I intentionally left some albums out because it becomes too hard to make them out in the graph when there's too many. (the graphing software I use is also not very inventive with its colors; it just starts reusing them).

Regarding TDEN, that section is ridiculously clipped. When you zoom into it, there's large swaths of samples at the waveform limit for long stretches of time.

I think the part I don't get is, when that happens in any kind of professional audio software, it throws red flags all over the place. They must have actively ignored those red flags to arrive at that level of compression.

[Calvin6s]

What about comparing it to some other album classics outside of Dream Theater?

[Calvin6s]

What genre was the first to really start this super hot, compressed sound?  I'm thinking industrial music?

[rumborak]

Problem is, I need the WAV files. For the DT albums I got the FLACs, but I would have to go out of my way to get it for other albums.

[Calvin6s]

Problem is, I need the WAV files. For the DT albums I got the FLACs, but I would have to go out of my way to get it for other albums.

My entire CD collection is in FLAC format, but I think it is illegal for me to share with you.   

Although your second graph is neat to check out, I liked the original graph which showed the flattening of the actual notes / drum hits / sounds.

I'm betting distorted guitar music in general is much less dynamic than other music.  Back in the day, I'd test audio equipment with something like DT's Scarred intro, simply because it was very dynamic compared to other metal albums.  Another favorite was The Cardigans Paralyzed
https://www.youtube.com/watch?v=zGVPsmwqh48
It has that nice low end, but also a lot of space.

There are test CDs, but you also need to kind of check how the equipment will color the music you listen to 90% of the time.

[erwinrafael]

rumbo, can I reqeust that you the same analysis with ToT. Just so I can clear in my mind that it is not really mostly a function of DT just playing louder more metal songs in the catalogue, than compression.

The new graph also now shows that SC is louder than DT12.

[Calvin6s]

The new graph also now shows that SC is louder than DT12.

But it also shows that it is softer.  And I always thought the problem wasn't necessarily how loud something is, but the range from soft to loud.

[erwinrafael]

Then it could really just be a function of having a variety of softer and louder songs. For example, SC has a long 10+ minute song that is not loud (Repentance), a long relatively quiet portion at the start of ITPOE 2, and the song portion of TMOLS is also relatively soft.

The reason that FII appears to have a lot of dynamic range in this graph maybe because it has lots of ballads (Anna Lee, Hollow Years, Take Away My Pain) but a lot of loud metal songs as well (Peruvian Skies, New Millennium, Burning My Soul).

Also, as the y axis appears to be absolute values instead of relative values, the curves are not really comparable. Although the shape could indeed show the distribution. But then again, the shape of the distribution could just be indicative of the turn towards more metal songs from early to current DT, not really of the effect of more compression in the mixes.

[Calvin6s]

The second graph definitely isn't a be all end all because of the factors mentioned.  More like a good overview.  The first graphs that show the amplitude going from sine wave like rises to flat cut offs is probably the more definitive view.  It suggests the highs were leveled so the lows can be raised, this squeezing (compressing) the range from low to high.  That's what it really all comes down to.

And in some ways that works for people that always listen via their mp3 player while working out or pumping the music in their cars.  I don't even bother listening to classical music in those situations can the lows can literally be drowned out by the surrounding environment for minutes.  And you have the option to crank it knowing you are going to have bleeding ears when the dynamics of the song rise again, you grin and bear it or you give up and skip to a different song (or song section).

There have been times where I've tried to listen to classical music in the car only to have to look at my dash or phone to make sure the music is still playing.  Just doesn't seem worth the effort sometimes.

[ariich]

Rumbo, in those graphs, just to check - are the SC and DT12 ones based on the CDs, or the HD tracks?

[hefdaddy42]

Then it could really just be a function of having a variety of softer and louder songs.

I don't think so.  And I have no idea why you would think that.

[rumborak]

Rumbo, in those graphs, just to check - are the SC and DT12 ones based on the CDs, or the HD tracks?

All CD tracks. I don't have the HDTracks. Would be a very interesting comparison indeed, but I'm not willing to dish out the money.

Regarding the graph, what I find it shows very well is how DT has increasingly succumbed to the loudness war. I think DT12's distribution shows exactly what people perceive: There's no middle ground anymore in terms of volume, it's full blast to the point of clipping most of the time.

[Calvin6s]

Is there a way to create a program routine that specifically counts the amount of times the amplitude gets hard limited like the graph that had the arrow pointing it out?  That could be a useful tool for those mixing their own songs.  Of course clipped at the source (microphone to *tape*) are hard coded mistakes.

[Skeever]

According to the more typical ways of measuring dynamic range, the vinyls and HD tracks versions of the last two albums are supposed to be good.

https://dr.loudness-war.info/album/list?artist=dream+theater&album=

DT really can't be unaware of the loudness war since it's been going on way too long. Maybe they simply feel that you want to hear the best version, you pay the premium :/

[BlobVanDam]

Even the DR value doesn't really give a perfect relative comparison. The HDTracks version of DT12 certainly sounds much better than the CD, but it does not sound comparable to a true DR12 as it indicates. There's a lot of headroom for the peaks of the drum hits, but the individual tracks are still very compressed.

Surely somewhere out there is a more accurate way to compare the volume of albums that is a little smarter.

[Calvin6s]

Most of the DR checking is just to reassure ourselves that we aren't taking crazy pills.  It seems like the backlash is building a bit more every year, but everything seems "big" on the internet.

The internet.  Invented for porn and bitching.

[BlobVanDam]

The DR is useful to indicate the negative trend in the industry, and it's a good approximation of the compression applied to an album, I just wish there was a bit of a more complete measure that was still quick and easy for people to read.
Perhaps a graph of DR over frequencies would be a good solution. Maybe that already exists.

[rumborak]

I think one of the difficulties with finding an automatic measure is, while say SC is heavily compressed, different sections are compressed to different levels. In any automatic measure this will show up as variation, but reality is that the ear easily tells the compression.

EDIT: Hmm, maybe a sliding window (of say a few seconds) would work, and the measure is the dB variance in that window.

[Stadler]

Most of FII looks like a rectangle too, but it still sounds pretty good. You have to use your ears aswell as your eyes.

Uh, call me old-fashioned, but it's MUSIC.  I'm pretty much going to use just my ears. 

[Stadler]

According to the more typical ways of measuring dynamic range, the vinyls and HD tracks versions of the last two albums are supposed to be good.

https://dr.loudness-war.info/album/list?artist=dream+theater&album=

DT really can't be unaware of the loudness war since it's been going on way too long. Maybe they simply feel that you want to hear the best version, you pay the premium :/

Or maybe they disagree with your opinion on the "best version". 

If you can't tell, not a freedom fighter in the loudness wars.   I don't think it is the whole story, by any means, but I think that some are overly minimizing the notion that artists - musicians - want their music heard.  And they want it to sound as good as it can to as many people as it can.  So there will invariably be some that either want it to sound a different way generally, or listen to their music differently than the majority.   I'm not at all understanding why musicians should forsake the majority of their audience in order to cater to a relatively small minority.

[ariich]

Then it could really just be a function of having a variety of softer and louder songs.

I don't think so.  And I have no idea why you would think that.

Well, actually that definitely would have an impact. I'm not suggesting for a second that it would explain the whole difference, and certainly in a case like the regular DT12 mastering where even the quietest moments (except perhaps the ambient/orchestral sections) are pushed right to limit it would be irrelevant. But where the production does allow for some dynamic range, then the musical balance between louder and quieter sections would affect the results.

[rumborak]

So there will invariably be some that either want it to sound a different way generally, or listen to their music differently than the majority.   I'm not at all understanding why musicians should forsake the majority of their audience in order to cater to a relatively small minority.

I think you have the implicit assumption there that DT are aware of the in-depth reception of their albums. I don't think that is the case; I think all they are exposed to are ultra-sugarcoated praises they get during Meet & Greets, and general sales numbers. In particular, DT only hears why people *still* listen to their music. They never hear why certain people *stopped* listening to their music.
So, my best guess is DT is operating somewhat in the dark, and thus go for what they perceive the "greater public" wants. This will certainly also be guided by RR's ideas on the matter.
Also, another thing to comment on, keep in mind that barely any artist listens to their own albums, i.e. the final product, more than twice. I don't think DT are aware of the low replay value of DT12. They will have listened to it once when they got the CD, concluded "this album is strong, and pushes hard", but never ended up trying to listen to it 10 times.

[PixelDream]

I remember quite clearly that Jordan said in an interview that Dream Theater were permanently 'done' with the whole loudness war. This was around (I think before) DT12 came out.

[BlobVanDam]

I remember quite clearly that Jordan said in an interview that Dream Theater were permanently 'done' with the whole loudness war. This was around (I think before) DT12 came out.

They also made comments about not wanting to be part of the loudness war at the time of BCASL too, so I wouldn't expect a major change in approach.

[tweeg]

That graph is incredibly interesting Rumborak. Thanks so much for taking the time to do it up for us.

There are a few things I notice, besides the ones previously discussed. Images is the quietest album by far, and it all sits in a nice little pocket with loads of (maybe too much) headroom.
Awake is similar, but louder, though there is a smaller plateau before the main hump. I suspect TSM and SDV were mastered almost separately from the rest of the album, and given little compression.
FII is quite loud, though there seems to be a fair bit of dynamics throughout. Its a common opinion that this is the best sounding loud DT album, and this graph lends credence to that.
SFAM is quite contrasted between its light and heavy songs, making a very dynamic album overall (the dynamics within each track notwithstanding).

I'd love to hear your thoughts on my opinions. Do they sound plausible?

[rumborak]

Totally agree on all of it, yeah. I think SFAM is particularly interesting in this sense, since it has a main hump not *that* far away from the main hump of the two most compressed, SC and DT12. I think the reason why it works is because SFAM has a lot of breathers between hard-hitting songs. Same with FII, but instead of having very quiet breathers, it rather has an extended slope to the softer parts.
I think FII probably stands out because it gets closed to the desired

------------\
                 \
                  \

shape.

[fischermasamune]

Shouldn't the area below the curves be the length of that album? IAW appears to have more area below than DT12, por example. Is this possibly due to the fact that the graph starts at 2000 and not 0 (which I imagine is the lowest RMS)?

[rumborak]

Yeah, I didn't show the area of 0-2000 because that area spikes up very high for all albums, way outside the range (or conversely, if one were to show the 0 range, the other stuff would be hard to tell).

[Kotowboy]

What genre was the first to really start this super hot, compressed sound?  I'm thinking industrial music?

Some people say it was the oasis debut Definitely Maybe - which was mixed to have no dynamics at all.