Article: Dynamic Range Day

[firedog]

I can agree with SG's stance to an extent, especially with modern, pop, hip-hop, rock - it is supposed to sound like that, so WE need to accept that or just not listen  to it. Complaining about it doesn't make much sense.

 

Where he's totally wrong: a) legacy material - why should e.g., a Springsteen album that had DR of 14 in the 80's be remastered to have DR of 6? It makes no sense, and claiming that the remaster VC is an "artistic" decision is a pretty weak one. And don't mention "hi-res" releases that this is done to; b) acoustic instruments - jazz quartets, etc. They are also starting to appear with volume compression to bring the DR from 16 to 8. Not unlistenable, but why? I'd guess even the people that buy them would prefer them without the added VR. And it adds NOTHING to the "art" - it merely reduces our ability to hear the subtleties of the playing. 

[JediJoker]

On 3/30/2017 at 6:24 PM, AlexMetalFi said:

Yet when I talk about an album sounding louder, I'm not talking about you listening to it at high volumes. Instead, I'm talking about records that have been made artificially loud by reducing their level of dynamic range, or the ratio between the loudest parts to the softest ones, during the mixing or mastering process. Very simply, an engineer will raise all the quietest parts of a recording by cutting off the low and high end bits regardless of whether or not the source material warranted it. This in effect squashes the original signal and in the process takes out a lot of the punch and sonic depth the recording would have had if this kind of aggressive limiting hadn't been applied. Take note, once a recording has gone through this process it is irrevocably damaged; there is no way to recover the samples that have been effectively chopped off, and more often than not, you are simply left with an amorphous wall of sound where volume homogeneity reigns supreme at the expense of ultimate fidelity.

 

I'm afraid there is some confusing misuse of jargon in this paragraph.  Dynamic range compression does not "cut off the low and high end bits," nor does it "effectively chop off samples."  It does reduce the number of bits needed to accurately represent the audio signal, but these bits are not removed from the file.  Likewise, while the content of samples is changed, there is no removal of samples.  If the file's coded sample rate did not change, removing enough samples would result in a very strange sounding bit of audio when played back!

 

It is also important to note that the "dynamic range" we're talking about is really closer to something called "crest factor" in signal theory; more simply, it is the "peak-to-loudness ratio."  It is the difference in level between the loudest and the average (average dBFS RMS or integrated loudness).  It is not the absolute dynamic range—the difference in level between the loudest and softest—which is, in most cases, the full available dynamic range afforded by the format (all available bits) due to "silence" at the beginning and/or end of the file and peaks close to 0dBFS.  The absolute dynamic range is, therefore, next to useless.  The peak-to-loudness ratio, however, gives a much better indication of how a file might sound in terms of perceived loudness, punchiness of transients, and so on.

[AlexMetalFi]

@JediJoker

 

No argument from me. I agree I am using a few terms loosely as this is an introductory piece.

 

With that said, I think you misread what I wrote or at least maybe my terminology is as I said, is a bit too high-level and can seem vague at times for those who want to dig deeper.

 

When I say "chopped off" I mean the waveform not the bit depth. So when an engineer is forced to hypercompress and limit, he or she may slam everything to 0dbFS which effectively "chops off" the peaks. The sample is still there, true, but the waveform has been damaged irrevocably ("clips").

 

Note that Audacity "Find Clipping" tool I think scans for three successful 0dbFS samples to constitute a clip by default (since music in general does not look like square wave forms). If you take what I wrote and watched the video, I think it is close enough for a gentle introduction.

 

Anyway, I will get into this more in future articles. 

[esldude]

2 hours ago, firedog said:

I can agree with SG's stance to an extent, especially with modern, pop, hip-hop, rock - it is supposed to sound like that, so WE need to accept that or just not listen  to it. Complaining about it doesn't make much sense.

 

Where he's totally wrong: a) legacy material - why should e.g., a Springsteen album that had DR of 14 in the 80's be remastered to have DR of 6? It makes no sense, and claiming that the remaster VC is an "artistic" decision is a pretty weak one. And don't mention "hi-res" releases that this is done to; b) acoustic instruments - jazz quartets, etc. They are also starting to appear with volume compression to bring the DR from 16 to 8. Not unlistenable, but why? I'd guess even the people that buy them would prefer them without the added VR. And it adds NOTHING to the "art" - it merely reduces our ability to hear the subtleties of the playing. 

I think the reissues pretty well indicate the idea it is an 'artistic' decision of modern times is at best a fig leaf of coverage.  And the leaf is mostly fallen to the wayside. Which also makes me think the idea it is the art of current music is also probably not true for the most part.

 

For all I care rap and hip hop can have DR0.  Rock, modern rock, sorry, but so far I only hear things that would be good were they not ruined by the excessive loudness.  Haven't heard one thing that could even possibly be better due to it being so loudly mastered. So I accept it, listen on youtube or other streaming and otherwise spend not dollar one on such horrid recording releases.

[JediJoker]

1 hour ago, AlexMetalFi said:

@JediJoker

 

No argument from me. I agree I am using a few terms loosely as this is an introductory piece.

 

With that said, I think you misread what I wrote or at least maybe my terminology is as I said, is a bit too high-level and can seem vague at times for those who want to dig deeper.

 

When I say "chopped off" I mean the waveform not the bit depth. So when an engineer is forced to hypercompress and limit, he or she may slam everything to 0dbFS which effectively "chops off" the peaks. The sample is still there, true, but the waveform has been damaged irrevocably ("clips").

 

I just think you might be better served to reserve using the words "bit" and "sample" when talking about technical aspects of digital audio.  If neophytes continue to read your article series, and you then use the terms in a more technical way, they could easily get confused based on the way you used them in this first article.  I can understand not wanting to introduce the concept of bits and samples right away, but I f you meant to say "chops off the peaks," why not just say that?  "Peak" is a pretty easily understood term, especially when looking at waveforms and demonstrated well in that video.  Or, if you don't want to introduce "peak" in the intro, something like "chops off the loudest parts" would work.

 

I do appreciate you taking the time to write the article and your commitment to an ongoing series, as dynamic range is my number one "cause" in audio both as a consumer and an engineer.  It's just that the loose use of terminology and incomplete explanation rubbed me the wrong way.

[AlexMetalFi]

 

6 minutes ago, JediJoker said:

 

I just think you might be better served to reserve using the words "bit" and "sample" when talking about technical aspects of digital audio.  If neophytes continue to read your article series, and you then use the terms in a more technical way, they could easily get confused based on the way you used them in this first article.  I can understand not wanting to introduce the concept of bits and samples right away, but I f you meant to say "chops off the peaks," why not just say that?  "Peak" is a pretty easily understood term, especially when looking at waveforms and demonstrated well in that video.  Or, if you don't want to introduce "peak" in the intro, something like "chops off the loudest parts" would work.

 

I do appreciate you taking the time to write the article and your commitment to an ongoing series, as dynamic range is my number one "cause" in audio both as a consumer and an engineer.  It's just that the loose use of terminology and incomplete explanation rubbed me the wrong way.

 

Got it and thanks for the constructive feedback! I do appreciate it. If you've read my other articles around the web, I think you will see I am usually pretty good with the lexicon I choose to describe dynamic range, loudness, etc.

 

I will ask Chris to make the change to "peaks" since I agree that is both) more accurate and b) inline with what I was trying to get across.

 

Again, thanks Jedi - I do think we are of one mind on this.

[MarkS]

I was always a big Led Zep fan and consistently wondered why I thought a ton of my digital LZ sounded so bad compared to the records that I recalled listening  to in high school on a system that probably cost 5% of my current one. This might explain it.  Bummer!

[MarkS]

I was always a big Led Zep fan and consistently wondered why I thought a ton of my digital LZ sounded so bad compared to the records that I recalled listening  to in high school on a system that probably cost 7% of my current one. This might explain it.  Bummer!

 

Any reason the database and Roon don't exactly match?

[BigGuy]

With regard to the Adele "25" CD having a DR of 5, what is interesting to me is that a rip of the vinyl release measures at a DR of about 10, twice that of the CD and actually a pleasure to listen to.  Interestingly, the person who did the rip  returned the vinyl because "it sounded awful" but this was before hearing the CD!  So as the case with the 6 Million Dollar Man... "We have the technology...".   Adele must care about her audiophile audience...or at least listeners with a turntable!

I too have been exploring earlier recordings by various artists and have noted that many have very high DRs.  I enjoy the Blues genre and after reading that ZZ Top started out doing very bluesy music, I got a copy of their "First Album" which has an average DR of ~15.  Not only is it recorded well but IMO is a terrific album.

 

[firedog]

1 hour ago, MarkS said:

I was always a big Led Zep fan and consistently wondered why I thought a ton of my digital LZ sounded so bad compared to the records that I recalled listening  to in high school on a system that probably cost 7% of my current one. This might explain it.  Bummer!

 

Any reason the database and Roon don't exactly match?

 

The hi-res LZ re masters I have from a couple of years ago have only moderate VC, not the extreme amounts we are often referring to here. Certainly don't sound bad, IMO.  

 

Don't understand your question about Roon. One of the weakenesses of Roon is that their database often doesn't include the digital download versions of releases. So slight differences in the timing of tracks or numbering (release on disc is two CDs, download is 16 tracks numbered sequentially) shows up as "don't match"

[AlexMetalFi]

1 hour ago, BigGuy said:

With regard to the Adele "25" CD having a DR of 5, what is interesting to me is that a rip of the vinyl release measures at a DR of about 10, twice that of the CD and actually a pleasure to listen to.  Interestingly, the person who did the rip  returned the vinyl because "it sounded awful" but this was before hearing the CD!  So as the case with the 6 Million Dollar Man... "We have the technology...".   Adele must care about her audiophile audience...or at least listeners with a turntable!

I too have been exploring earlier recordings by various artists and have noted that many have very high DRs.  I enjoy the Blues genre and after reading that ZZ Top started out doing very bluesy music, I got a copy of their "First Album" which has an average DR of ~15.  Not only is it recorded well but IMO is a terrific album.

 

 

Vinyl DR scores are bogus. More than likely it was the same master and level matching was not performed. That would be my guess.

[wwaldmanfan]

4 minutes ago, AlexMetalFi said:

 

Vinyl DR scores are bogus. More than likely it was the same master and level matching was not performed. That would be my guess.

I have numerous examples in my library where the vinyl sounds excellent, with full dynamics, and the digital download/CD version of the same album is compressed junk.

[AlexMetalFi]

1 hour ago, wwaldmanfan said:

I have numerous examples in my library where the vinyl sounds excellent, with full dynamics, and the digital download/CD version of the same album is compressed junk.

 

So do I. But you have to verify that they were sourced from two different masters.

 

More often than not, the CD master is pressed on the vinyl as is (or with some minor EQ to get it to cut).

[astrotoy]

On 3/31/2017 at 10:36 AM, AlexMetalFi said:

 

Follow the directions for Offline Meter here:

 

http://www.metal-fi.com/measuring-dynamic-range/

Thanks.  I was able to do DR for several of my CD's.  Is there a stand alone program that will do DR for higher rez files (I have lots of 192/24 rips from vinyl and tapes, and quite a few DSD files)?   Larry

[PeterG]

2 hours ago, AlexMetalFi said:

 

Vinyl DR scores are bogus. More than likely it was the same master and level matching was not performed. That would be my guess.

 

2 hours ago, wwaldmanfan said:

I have numerous examples in my library where the vinyl sounds excellent, with full dynamics, and the digital download/CD version of the same album is compressed junk.

 

Alex,

Please explain why vinyl DR scores are bogus.  Do you mean that the posted measurements are inaccurate or misleading?  I am all digital today, but I've been thinking about adding a turntable in the hope of matching wwaldmanfan's experience.  But if I cant trust the posted scores...

Thanks

[AlexMetalFi]

 

Bottom line: Just because a vinyl needle drop will score DR11 on the TT, doesn't mean squat. I HAVE in front of me a needle drop of a record and its CD release  One is DR11 the other is DR6. And guess what? They were both sourced from the original master. (I actually have the original undithered 24-bit master as well sent to me by the engineer with permission from the label/band).

 

Of course, there are plenty of releases where the vinyl master does indeed sound better because it wasn't crushed. But don't stare at DR scores to tell you that - you need to level-match and listen.

 

I'll talk more about it in a future article to be sure.

[PeterG]

2 hours ago, AlexMetalFi said:

 

Bottom line: Just because a vinyl needle drop will score DR11 on the TT, doesn't mean squat. I HAVE in front of me a needle drop of a record and its CD release  One is DR11 the other is DR6. And guess what? They were both sourced from the original master. (I actually have the original undithered 24-bit master as well sent to me by the engineer with permission from the label/band).

 

Of course, there are plenty of releases where the vinyl master does indeed sound better because it wasn't crushed. But don't stare at DR scores to tell you that - you need to level-match and listen.

 

I'll talk more about it in a future article to be sure.

Thanks for the excellent video--I will watch others from him as well.  One possibility, that Ian seems to suggest around the 11 minute mark, is that vinyl is both lower fidelity and more pleasing.  Perhaps one of the analog elements he mentions introduces new peaks and those new peaks hide the unpleasantness of the brick wall effect in the original file?  

[The Computer Audiophile]

11 hours ago, MarkS said:

I was always a big Led Zep fan and consistently wondered why I thought a ton of my digital LZ sounded so bad compared to the records that I recalled listening  to in high school on a system that probably cost 5% of my current one. This might explain it.  Bummer!

 

Perhaps something else could explain why the Zeppelin albums sounded so good in the 70s :~)

 

Only joking. 

[santar]

10 hours ago, AlexMetalFi said:

 

Vinyl DR scores are bogus. More than likely it was the same master and level matching was not performed. That would be my guess.

 

8 hours ago, PeterG said:

 

 

Alex,

Please explain why vinyl DR scores are bogus.  Do you mean that the posted measurements are inaccurate or misleading?  I am all digital today, but I've been thinking about adding a turntable in the hope of matching wwaldmanfan's experience.  But if I cant trust the posted scores...

Thanks

Yes, this is worrying. Do you mean we cannot trust the DR scores of any vinyl rips? Please clarify.

[esldude]

Here is just a supposition.  Not yet tested out, but someone with a vinyl rip could check it easily enough.

 

Is the difference that vinyl has significant low frequency wobble or wow that artificially increases the DR measured.  Take a look at the screen shot.  One is the waveform view of a DR4 track.  The other is the same track with a -20 db 3 hz added to it. See how it modulates the flat top wave shape to something like the vinyl looks in the video above?  Most obvious is the tail in what was previously silence at the end.

 

Notice how this view of a vinyl rip middle of this page has a similar thick tail of silence at the end.  Lower level than my example.  Still all that is needed is for someone with a vinyl rip handy to see if noise in the single digit hertz range is present in the groove between tracks.

 

https://www.metal-fi.com/forum/viewtopic.php?t=186

 

 

 

 

 

[esldude]

It also occurs to me you could take the vinyl file and do a brickwall high pass filter at 30 hz and see if the DR value starts to approach that of the CD.  That would let you know it is being modulated by some low frequency speed variation.

 

[JediJoker]

22 hours ago, astrotoy said:

Thanks.  I was able to do DR for several of my CD's.  Is there a stand alone program that will do DR for higher rez files (I have lots of 192/24 rips from vinyl and tapes, and quite a few DSD files)?   Larry

 

My suggestion is to use the DR plugin for foobar.  It will measure anything you throw at it.

[Bullwinkle]

On 3/31/2017 at 2:09 PM, esldude said:

...I have done this for friends I recorded.  Add a bit of compression maybe a touch of reverb.  Let them hear it and the original.  Compressed is preferred.  Add a bit more compression and compare the two compressed versions where more compressed is preferred by them.  Rinse repeat a total of about 6 times.  Then play them the first compressed version and the last most compressed version.  Noses wrinkle, eyes squint, then they say something like what happened to that one.  The least compressed version is preferred and the highly compressed version in comparison sounds 'wrong'.  I wonder how often this accidentally happens and the last comparison never takes place side by side.  Your ear always prefers the slightly louder sound, and compression raises average loudness.  If comparing two already very compressed versions you still may go for the louder even at the point it is very messed up because both are messed up...

 

FYI, there's a very famous paper "Intransitivity of Preferences" published in 1969 in Psychological Review by the late great Amos Tversky that does this very same trick. Tversky called the perceptual structure leading to this type of cycle a lexicographical semiorder. I think I recall Tversky saying that the earliest reported observations of such cycles were for some kind of auditory stimulus too--perhaps varying loudness. I will look it up.

[astrotoy]

2 hours ago, Bullwinkle said:

 

FYI, there's a very famous paper "Intransitivity of Preferences" published in 1969 in Psychological Review by the late great Amos Tversky that does this very same trick. Tversky called the perceptual structure leading to this type of cycle a lexicographical semiorder. I think I recall Tversky saying that the earliest reported observations of such cycles were for some kind of auditory stimulus too--perhaps varying loudness. I will look it up.

 

Tversky's colleague Daniel Kahneman's fun book "Thinking Fast and Slow" is a good read for the layperson.  Talks about many issues which, I think, apply to audiophile nervosa.  Kahneman says several times that Tversky would have shared the Nobel prize with him, if he had not died prematurely.

Larry

 

 

[astrotoy]

6 hours ago, JediJoker said:

 

My suggestion is to use the DR plugin for foobar.  It will measure anything you throw at it.

Thanks.  I'll have to figure out how to get Foobar.   Larry