Dynamic range scores: a cautionary note

[Satanica]

Hi Satanica. Originally I suspected vinyl; now I suspect both. Refer post #128. cheers

 

Hey, I've noticed for some time that the latest version of JRiver calculates and displays a Dynamic Range R128 value after audio has been analysed. See here: http://yabb.jriver.com/interact/index.php?topic=82025.0

 

The values calculated of Daft Punk's Random Access Memories tracks are very similar when comparing digital vs vinyl (as we have discussed before). And of course as you know the DR values are significantly different.

 

Cheers.

Edited February 9, 2014 by Satanica

[davewantsmoore]

Hi Satanica. Originally I suspected vinyl; now I suspect both. Refer post #128. cheers

 

Ah, I missed that post... it's what I was getting at in #149

 

 

 

Oversimplification is the devil   ;-)

[davewantsmoore]

Hey, I've noticed for some time that the latest version of JRiver calculates and displays a Dynamic Range R128 value after audio has been analysed. See here: http://yabb.jriver.com/interact/index.php?topic=82025.0

 

The values calculated of Daft Punk's Random Access Memories tracks are very similar when comparing digital vs vinyl (as we have discussed before). And of course as you know the DR values are significantly different.

 

Yes, the EBU standard is better ...   the DR database really measures 'loudness', not 'dynamics'.

[davewantsmoore]

Oversimplification is the devil   ;-)

 

... and even I'm doing it because it doesn't even really measure what we hear as 'loudness'   (although it would be a more fitting description than 'dynamics') .....  it measures the peak to average ratio of a tiny sampling of the worst part of the track.

 

EBUR128 is a better measure of 'loudness'

[Newman]

the latest version of JRiver calculates and displays a Dynamic Range R128 value after audio has been analysed. See here: http://yabb.jriver.com/interact/index.php?topic=82025.0

 

The values calculated of Daft Punk's Random Access Memories tracks are very similar when comparing digital vs vinyl (as we have discussed before). And of course as you know the DR values are significantly different.

 

Hi, I don't fully grasp the jriver implementation, but here is what I think it is doing. (Maybe others can chip in.)

 

The R128 value should be a loudness measure, in LU's (Loudness Units). It is not in any way a measure of the dynamic range of the music.

 

I think the jriver term "Dynamic Range R128" is simply the LRA (Loudness RAnge) as defined in the Recommendation 128 documents. If so, it is the difference in LU (Loudness Units) between the 10th and 95th percentiles of the distribution of loudness levels in the song -- after gating out all data that appears to not be music (below a certain threshold). IMHO this measure could easily mislead with a high score for songs that have been dynamically compressed in the studio. It is what it says it is: a standard measure of loudness range.

 

The TTDR measure is an admirable attempt to do it right, and correlate a number with the dynamics of the music. I was pretty excited about it as a useful tool for audiophiles who need insight into the mastering quality of various issues of favourite albums, but investigations have shown that it doesn't work for this purpose.      

 

Right now I am not aware of any simple numeric or graphical display that gives us a good idea on this point.

 

[edit: spelling]

Edited April 10, 2014 by Newman

[Linkin]

In my opinion, before you can compare DR values between CD and Vinyl, you need to:

 

 

Digitise the vinyl, remove any pops/clicks, and then amplify it to match the level of the CD (easily achievable with audacity.) Then listen to both versions, make your analysis based off of hearing only & write it down. Measure second, and see if the measurement matches your notes.

 

 

A caveat is that the waveform can tell you a lot before you listen, on the macro side of things. You still actually have to listen to compare micro things like clarity and detail.

 

 

A few things to note:

 

 

I've found it's quite common for CD masters to have clipping, even if they haven't been brickwalled. I've also seen this on extremely popular classics, like dark side of the moon. Clipping is extremely worse on CD's that have been "maximised" - and you'll find that they all have a low dynamic range measurement as well.

 

 

Initial digitised vinyl can either be quiet, or be loud and have clipping. You need to lower soundcard input volume if it's clipping. I had to do this when I upgraded the stylus on my Stanton 681. The new needle had a much higher output! So it's a lot more fiddly than digital, but you can control volume/loudness, and also have a copy of your songs that don't clip.

 

 

Below are a few examples of CD's with clipping, and my vinyl versions after click removal/amplification

 

 

Dire Straits - Telegraph Road (this CD is a recent acquisition - first west german pressing). These measurements are per track not per album. CD version measures DR 13. Vinyl version measures DR 12. I'm 99.9% certain that my measurements are accurate here

 

CD: http://i.imgur.com/0gk6rwk.png

 

Vinyl: http://i.imgur.com/BDNHVl3.png

 

 

Pink Floyd - Money. CD measures DR 8. Vinyl measures DR 12 (It measures this amped or not amped) - For the record I don't own this CD, I borrowed it from my old man. Not sure what kind of pressing it is, but the whole album measures lower and clips and some songs (not all, Money is the best example here). Also noticed the "flat" peaks on the left channel, where the right is clipping

 

CD: http://i.imgur.com/SaUScbp.png

 

Vinyl: http://i.imgur.com/fFkMpv4.png

 

 

 

Holy mother of clipping, batman! Please, open a few tracks in audacity and turn "show clipping" on. You'll be horrified.

Edited February 14, 2014 by Linkin

[Newman]

Hi Linkin. That is actually a very mild example of clipping. Also, a small amount of clipping does little to no harm for this type of music. My suggestion is to look at the waveform much more closely on an expanded time scale, before concluding.

 

My other suggestion is to look at your phono cartridge. Its right channel output is significantly lower than the left. Could be setup, or more likely the cart itself. Or even the phono stage.

 

Enjoy your journey. Try not to get too excited about whole-song waveform plots. Ignore DR if you already have both versions of an album, just pick the one you want to listen to the most!  

[Linkin]

I have much worse examples of clipping... though I'd prefer it if CD didn't clip at all

[Newman]

Me too.

[Newman]

Here is a video that shows that you cannot use the DR scores for vinyl for comparisons with digital editions of the same music. It is an exact duplicate example of the example I raised with the Daft Punk RAM album, with the exact same outcome as I raised with RAM.

 

 

Conclusion: a recording with no more dynamics on vinyl, will routinely measure a higher DR score than its digital equivalent, by an unpredictable amount. And audiophiles the world around will claim their ears have confirmed the extra dynamics in the vinyl. Like you did with RAM, millsy, on another thread.

 

And therefore, to repeat, it is best to ignore the relative DR scores of digital and vinyl. And the comparative whole-song waveform plots.

 

So, nothing new here in terms of material for this thread. But the validation by a third party is nice.

 

A bit more validation, on the archimago blog entry last week:

 

I noticed this afternoon playing with Audition just how easy it was to artificially inflate DR values! For example, mixing all low frequencies <100Hz as mono. I played around with this using Lorde's "Royals" song using a lowpass filter to isolate those low frequencies and then mix them back into the highpassed upper portions as mono... Even though the sound wasn't significantly different when volume matched, it was not difficult at all to get a DR8 original 24/48 HDTracks version up to DR13.

So, he mixed the bass to mono and DR score jumped 5 points (DR8 to DR13). Guess what? Mixing the bass to mono is universal practice for preparing a vinyl mix.

 

This 'discovery' made him recall the post he put up last week, where he wrote about using DR scores as an indication of different masters. Wise move!

[Linkin]

A DR measurement where the CD measures 12 and the vinyl measures 24 is flawed.

 

A DR measurement where the CD measures 12 and the vinyl measures 13 or 14 is still flawed, but it's most likely that they use the same studio master (hopefully not remastered CD mastered to vinyl - a la 2008 metallica re-releases). Looking at the waveforms should also tell the same story.

 

Now if you're comparing DR values from the same source (CD, or Vinyl, not one against the other) then it's useful for meaningful measurements.

Edited March 17, 2014 by Linkin

[Newman]

It's such a mess that we can't even draw those conclusions, I'm afraid.  Like archimago concluded in the blog post I linked above, "Given the effect I saw, ultimately I think there's no conclusion one can draw between what's measured on a vinyl rip and knowledge/proof of whether truly new masters are being used unless specifically told about it!"

 

That's about right!

[Sentient]

Looking at the waveforms should also tell the same story.

.

Exactly.

There was a useful thread that did exactly that.

Can't believe Newman is still campaigning against DR ratings. It's almost like you prefer compressed music??

[Sentient]

Like archimago concluded in the blog post I linked above, "Given the effect I saw, ultimately I think there's no conclusion one can draw between what's measured on a vinyl rip and knowledge/proof of whether truly new masters are being used unless specifically told about it!"

Tell him to use his ears, it's blatantly obvious.

[Newman]

Troll day today, my friend?
 
You haven't been in this thread since August, and your last comments to me on this general topic went like this:
 

Newman,
Can we PLEASE agree to disagree on this discussion. I will give you a wide birth, if you give me, and this thread, a wide birth.

Continually badgering me and my posts, again and again, over and over, with the same argument is really tiresome.

Well, ditto. Unless you wish to make a retraction, and resume discourse? Have you got anything new? Not 'badgering', of course, which is off limits by agreement.

 

Posts like #160, above, are not me 'campaigning'. It is good solid information by a third party, demonstrating just how easily the TTDR meter is tricked in relation to dynamic range of the recording itself. Nobody has rationally debated it, because ....?

 

And that is why this thread settled down nearly 2 months ago.

[Sentient]

Petty insult cast aside.

 

Posts like #160, above, are not me 'campaigning'. It is good solid information by a third party, demonstrating just how easily the TTDR meter is tricked in relation to dynamic range of the recording itself. Nobody has rationally debated it, because ....?

Any tool can be "tricked" when you know how the tool works.

At the end of the day, it's how it sounds. Now we can argue until we are blue in the face, that records sound different due to the playback system (and all the other variables in producing them). Sure. That is definitely in play.

 

We can question whether the record is from the same post production digital master, or it isnt,,,, does it really matter?

What matters is how it sounds, the end result. What do you prefer?

It's widely known that records are mastered by a different mastering engineer, to the digital production (unless it is someone special like Doug Sax).

The question is, what is the vinyl mastering engineer given?

It's considered "best practise" to give the finalised stereo mix from the studio, to the vinyl mastering engineer. But not everyone follows the same path.

Path1: Finalised Studio Stereo Mix -> Digital Mastering Engineer -> [CD pressing plant / Vinyl mastering,pressing]

Path2: Finalised Studio Stereo Mix -> Vinyl Mastering Engineer -> Vinyl pressing

Path 1 or 2 is a conscious choice. You cant say it is always the same path. That just isnt factual.

Can we definitely say an album follows path 1 or 2, based on DR vales alone. No, I'm not suggesting that.  

Instead of quoting 3rd parties on other forums, how about creating some real SNA generated analysis?

Here are some 30sec samples. Listen to them. They sound very different, even on cheap computer headphones. I have volume matched them around the voice, as close as I can:

Right Click (Save-As)- Sample1

Right Click (Save As)- Sample2

Not everyone will prefer the same sample. I expect alot of people to prefer the compressed one, because it is heavier in bass and warmer as a result. But it lacks life, it lacks air. On a genuine hifi system (balanced evenly from top/bottom) they are polls apart. I know which I prefer (and yes, the DR ratings do support my preference).

Edited May 6, 2014 by ozmillsy

[Newman]

If I knew what the point was, I might show some interest. Does it disprove the evidence that TTDR is misleading? How?

 

If it is about CD vs vinyl masters, please move it to another thread. See here and here and here, etc. 

[hochopeper]

Any tool can be "tricked" when you know how the tool works.

 

Isn't the point actually that if you don't know how the tool works you don't know if it is being applied properly and potentially giving a misleading result.

[Sentient]

If I knew what the point was, I might show some interest. Does it disprove the evidence that TTDR is misleading? How?

In this thread, you've clearly stated that vinyl DR ratings are flawed, and you continue to push that view.

My point is, it's how the music sounds. Form a view based on what you hear, first and foremost.

Then look at other things like waveforms, DR ratings, etc. Do these measures support what you hear or dont they? If they dont, then they are misleading. Simple.

In my experience the DR ratings I see support how it sounds (on my system). They're not misleading at all.

[Sentient]

Isn't the point actually that if you don't know how the tool works you don't know if it is being applied properly and potentially giving a misleading result.

No, that isnt the point I'm making. Anyone can manipulate numbers to prove a theory, or discredit results.

It's not about whether you know how the tool works. It's whether you trust the results it gives you.

Newman is suggesting we shouldnt trust the results for a whole bunch of reasons, or at the very least be cautious about the values you're getting.

I'm suggesting, let your ears be the judge on whether the DR rating results are consistent with how the music sounds.

Edited May 6, 2014 by ozmillsy

[hochopeper]

Sounds like Newman's the guy that's done the statistics course:

 

 

 

Just because your ears and the tool might agree, doesn't make the tool fit for that application.

Edited May 6, 2014 by hochopeper

[Guest Muon]

Please don't take our saviours name in vane, thank you.

:blink:

[Sentient]

Just because your ears and the tool might agree, doesn't make the tool fit for that application.

LOL, it's a start.

The DR ratings are just an averaging formula, to assess the amount of dynamic range between different tracks/music. It's not the be all and end all. It's just a measure to be used in context.

The fact that people even consider and discuss dynamic range at all, is a step in the right direction, IMHO. I dont care what tools they use. Trying to discredit the tool, is like cutting off your nose to spite your face.

[Sentient]

{innane drivel snipped}

The right thing to do is to expose and warn against the broken tool, use your ears just like always, and encourage the search for a tool that isn't broken. That's my position.

Please provide evidence that the TTDR tool generates invalid results.

Give me a sample track of real (published) music that hasnt been doctored, and has dynamics in place. Show me the incorrect DR results.

[Satanica]

I'm not sure if anyone has posted a visual representation of the waveforms of Daft Punk's RAM but here is the first track from Audacity. As we know the Vinyl version has higher DR values.

 

Now if you compare the peaks of the vinyl they look natural i.e. the drummer sometimes hitting stuff louder than at other times. But if you look at the HD Tracks version (apparently the master) you'll notice they're basically all as loud as each other as if they've been put through some sort of compressor/limiter.

 

Just for shits and giggles I "zoomed in" on the Vinyl version, I think you'll notice it looks incredibly like the HD Tracks version now.

Edited May 25, 2014 by Satanica