I don't understand the obsession with DR meters

[Sal1950]

I have to disagree - not in principle, but empirically. The Loudness Wars seem to have let up a bit, but there are still way, way too many modern masters of all kinds of music that are mastered too loud.

Thank goodness I don't listen to a lot of modern masterings. 90-95% of what I listen to was done before all this craziness started and a lot of the newer remasters I do listen to was done by engineers who wouldn't butcher music like that. I just checked my latest acquisition, Blackfield's V in LPCM 24/96 5.1 ch master by Steven Wilson. Still showing a nice DR11 average with no track under 2 digits.

foobar2000 1.5.3 / Dynamic Range Meter 1.1.1
log date: 2020-07-05 12:35:06

--------------------------------------------------------------------------------
Analyzed: Blackfield / Blackfield V (MCH BD LPCM 96k 5.1)
--------------------------------------------------------------------------------

DR Peak RMS Duration Track
--------------------------------------------------------------------------------
DR10 -10.65 dB -28.30 dB 1:23 01-A Drop In The Ocean
DR11 -0.10 dB -13.47 dB 3:37 02-Family Man
DR12 -0.10 dB -16.40 dB 3:59 03-How Was Your Ride
DR12 -0.10 dB -14.95 dB 2:54 04-We'll Never Be Apart
DR11 -3.87 dB -20.25 dB 3:05 05-Sorrys
DR12 -0.10 dB -16.86 dB 3:22 06-Life Is An Ocean
DR11 -0.10 dB -15.03 dB 3:23 07-Lately
DR10 -1.87 dB -17.45 dB 3:31 08-October
DR12 -0.10 dB -16.11 dB 3:57 09-The Jackal
DR13 -0.10 dB -16.50 dB 2:39 10-Salt Water
DR11 -0.10 dB -14.05 dB 4:02 11-Undercover Heart
DR13 -0.10 dB -15.66 dB 3:42 12-Lonely Soul
DR11 -0.10 dB -14.20 dB 4:31 13-From 44 To 48
--------------------------------------------------------------------------------

Number of tracks: 13
Official DR value: DR11

Samplerate: 96000 Hz
Channels: 6
Bits per sample: 24
Bitrate: 7591 kbps
Codec: FLAC
================================================================================

 

[maverickronin]

Dynamic Range Meter 1.1.1

It seems like that plugin has half dissapeared from the internet. I found a random dropbox link and ran it though virustotal so here's a copy for anyone else who wants it....

Also, random EP from from a group whose whole discography I bought yesterday.

foobar2000 1.3.8 / Dynamic Range Meter 1.1.1
log date: 2020-07-05 12:12:28

--------------------------------------------------------------------------------
Analyzed: NECRONOMIDOL / DAWNSLAYER
--------------------------------------------------------------------------------

DR         Peak         RMS     Duration Track
--------------------------------------------------------------------------------
DR3       -0.02 dB    -4.64 dB      4:04 01-DAWNSLAYER
DR5       -0.02 dB    -6.56 dB      4:21 02-STARRY WISDOM
DR3       -0.01 dB    -4.37 dB      5:34 03-R'LYEH
DR6       -0.02 dB    -8.74 dB      4:03 04-celephaïs
DR3       -0.02 dB    -4.82 dB      3:16 05-ABHOTH
--------------------------------------------------------------------------------

Number of tracks:  5
Official DR value: DR4

Samplerate:        44100 Hz
Channels:          2
Bits per sample:   16
Bitrate:           1130 kbps
Codec:             FLAC
================================================================================

I can't really say any of those "low" numbers bother me as it sounds great on my Stax...

 

[Sal1950]

I can't really say any of those "low" numbers bother me as it sounds great on my Stax...

What would most likely bother me is the style of music itself, if you enjoy listening to that I don't imagine the DR would matter at all, that's the way it was designed to sound. I'm not throwing stones, just something I see as fact, a limited DR gives the listener the effect the artist intended.

 

[ernestcarl]

It seems like that plugin has half dissapeared from the internet. I found a random dropbox link and ran it though virustotal so here's a copy for anyone else who wants it....

Also, random EP from from a group whose whole discography I bought yesterday.

foobar2000 1.3.8 / Dynamic Range Meter 1.1.1
log date: 2020-07-05 12:12:28

--------------------------------------------------------------------------------
Analyzed: NECRONOMIDOL / DAWNSLAYER
--------------------------------------------------------------------------------

DR         Peak         RMS     Duration Track
--------------------------------------------------------------------------------
DR3       -0.02 dB    -4.64 dB      4:04 01-DAWNSLAYER
DR5       -0.02 dB    -6.56 dB      4:21 02-STARRY WISDOM
DR3       -0.01 dB    -4.37 dB      5:34 03-R'LYEH
DR6       -0.02 dB    -8.74 dB      4:03 04-celephaïs
DR3       -0.02 dB    -4.82 dB      3:16 05-ABHOTH
--------------------------------------------------------------------------------

Number of tracks:  5
Official DR value: DR4

Samplerate:        44100 Hz
Channels:          2
Bits per sample:   16
Bitrate:           1130 kbps
Codec:             FLAC
================================================================================

I can't really say any of those "low" numbers bother me as it sounds great on my Stax...

I know those are headphones, but what SPLs do you usually listen at? Don't you find yourself lowering the volume knob much, much more as compared to more dynamic albums?

 

[ernestcarl]

What would most likely bother me is the style of music itself, if you enjoy listening to that I don't imagine the DR would matter at all, that's the way it was designed to sound. I'm not throwing stones, just something I see as fact, a limited DR gives the listener the effect the artist intended.

Just wondering, have you tried listening to Skrillex? Modern stuff, I know, I know... but it can be fun when taken in small doses.

 

[maverickronin]

I know those are headphones, but what SPLs do you usually listen at?

Usually 60-70 if I'm going to be listening for several hours, 80-90dB if I'm just going to be listening to one album, and I occasionally crank it to 100+ for a single track.

Don't you find yourself lowering the volume knob much, much more as compared to more dynamic albums?

Not that much. I end up setting the volume more by the peaks than the average level. If the peaks are rare enough, like with some classical, then I end up manually "gain riding" with the volume control. (Which is one of the reasons I rarely listen to classical even though I actually like a lot of it...)

 

[maverickronin]

I just checked my latest acquisition, Blackfield's V in LPCM 24/96 5.1 ch master by Steven Wilson.

I checked out the 2.0 version on Spotify. I wouldn't change the station if it was playing on the radio.

 

[ernestcarl]

I end up setting the volume more by the peaks than the average level. If the peaks are rare enough, like with some classical, then I end up manually "gain riding" with the volume control. (Which is one of the reasons I rarely listen to classical even though I actually like a lot of it...)

I'm guessing you probably don't listen to whole classical albums as opposed to more mixed classical playlists? Most of the time, I don't change the volume once set for any given album. Watching an opera on blu-ray, for example, I don't find myself reaching for the volume control due to loud or quiet passages -- maybe grab the remote for the pause button for the inevitable "pee break", as such performances can be extremely long -- LOL.

 

[maverickronin]

I don't change the volume once set for any given album. Watching an opera on blu-ray, for example, I don't find myself reaching for the volume control due to loud or quiet passages

That's exactly what I do when I try and listen to a whole symphony or something.

I'm guessing you probably don't listen to whole classical albums as opposed to more mixed classical playlists?

I don't even have any, although that would probably work better for me. It's probably about time for me to try to get into classical again as I know there's a ton of great stuff I'm missing, but it has such a high barrier to entry.

 

[Sal1950]

I wouldn't change the station if it was playing on the radio.

So you enjoyed it then, good.
There's hope for you yet.

 

[Sal1950]

Just wondering, have you tried listening to Skrillex? Modern stuff, I know, I know... but it can be fun when taken in small doses.

I have a very hard time calling something made without musicians playing musical instruments "music".
OK to dance to I guess but at 70 I'm a little too old to dance to that, and otherwise painful for just "listening" for more than a minute or two.
YMMV

 

[bennetng]

It seems like that plugin has half dissapeared from the internet.

Reason:
https://help.foobar2000.org/troubleshooter/components/27209dfe26d9450f
For example, if you use this plugin to scan tracker formats and MIDI files, foobar will crash.

Anyway this plugin is a joke. If you have a cassette deck, you can try this experiment as well:
https://www.audiosciencereview.com/...-dragon-cassette-deck.5595/page-2#post-124666

The files above scanned with the DR plugin:
https://www.audiosciencereview.com/forum/index.php?threads/high-resolution-audio.6525/post-190684

 

[maverickronin]

Reason:
https://help.foobar2000.org/troubleshooter/components/27209dfe26d9450f
For example, if you use this plugin to scan tracker formats and MIDI files, foobar will crash.

Yeah, I saw that, but it's not really a big deal. You know when you're using it, so you should prepared for it crash if it decides to.

Anyway this plugin is a joke. If you have a cassette deck, you can try this experiment as well:
https://www.audiosciencereview.com/...-dragon-cassette-deck.5595/page-2#post-124666

The files above scanned with the DR plugin:
https://www.audiosciencereview.com/forum/index.php?threads/high-resolution-audio.6525/post-190684

My understanding was that the whole "DR" algorithm itself is a joke since it can be manipulated so easily and it's not just this plugin's implementation of it. Is that not correct?

I tracked it down since this thread is more or less about how "DR" ratings relate to quality.

 

[bennetng]

the whole "DR" algorithm itself is a joke since it can be manipulated so easily

Yes, that's what I think.

 

[tmtomh]

Yeah, I saw that, but it's not really a big deal. You know when you're using it, so you should prepared for it crash if it decides to.

My understanding was that the whole "DR" algorithm itself is a joke since it can be manipulated so easily and it's not just this plugin's implementation of it. Is that not correct?

I tracked it down since this thread is more or less about how "DR" ratings relate to quality.

The DR algorithm is simple: DR = 2nd highest peak minus RMS. Because it is so simple, it can of course be "gamed," and there are situations where it is not useful at measuring or comparing actual dynamics. Most notably, as the cassette link you provided shows, the DR meter is not reliable for digital-analogue-digital chains, aka vinyl or cassette or reel-to-reel "rips ", because the distortions and inaccuracies the analogue step introduces inevitably randomize the peaks, softening peak-limiting and introducing new, spurious peaks that make the waveform look better and artificially raise the DR value. And it is certainly true that the DR meter has caused way too many people to think that vinyl is "more dynamic" than digital. The truth is that as a format it is not - and in addition, some of the mastering moves made to prep a source for vinyl cutting tend to artificially boost DR (see below). On the other hand, some vinyl masterings are done by different engineers, and as they targeted for an "audiophile" audience, some vinyl versions are in fact more dynamically mastered than their digital counterparts.

But the main point, as I see it, is that using the DR meter in an application for which it was never intended (measuring the DR of digital rips of analogue media) does not prove the DR algorithm is a joke. It still has reliable predictive and comparative power.

For example, the errors that vinyl (and cassette) rips create are remarkably consistent, despite the huge variation in vinyl mastering, vinyl pressing, and the rippers' playback equipment. The DR meter misreads the dynamic range of such rips compared to digital sources of the same material - but it consistently misreads it in more or less the same way, to more or less the same degree.

Now, there are still other issues, I admit: a bass-shy or overly trebly digital mastering will tend to read higher on the DR meter despite not being more dynamic in any meaningful way. And a CD made from an LP master tape that has narrowed or mono'd L-R bass also will tend to read slightly higher on the DR meter. But these will generally produce differences of 1-2dB on the meter, which are not terribly significant.

So IMHO the DR meter is a limited tool, but still a valuable one. The difference between a DR6 mastering of a digital source and a DR9, 10, or 11 mastering of that same digital source are not subtle, and I've never yet run into a situation where the latter did not sound better than the former.

 

[bennetng]

To further elaborate, the algorithm determines DR value by looking at the distribution of amplitude sample values. If a lot of samples fall into a narrow range of amplitude values, then DR will be lower. Try the attached test files:

By listening to the files, they sound the same to me, RG also showed they have identical loudness (suggested track gain based on -18LUFS target loudness), just different peak values, but the DR meter showed 9dB differences.

Among other factors (e.g. physical and mechanical properties of vinyl disc, turntable etc), RIAA and inverse RIAA filters in vinyl unavoidably change the phase and frequency response, and therefore also changed the distribution of amplitude values in the rip. The same applies to analog tape and cassette as well. They don't use RIAA filters, but they also have frequency response and phase issues. The differences and magnitude of differences shown on the DR meter are not a proof of different mastering, as explained in the famous Ian Shepherd video illustration.

 

[wgb113]

Age and music preferences are definitely a factor but to me the song matters most. I'd rather listen to a good song that's DR4 than some audiophile show music that a DR12. Some combinations of the two is obviously ideal.

My kids - 14 and 7 - find anything with wide dynamics very jarring like the OP. Whether it's a song or a show/movie they've grown up in a time where current releases can be squashed and normalized to death. The impact of soft/loud passages is completely lost on them and found to be an annoyance.

 

[bennetng]

Just another funny behaviour of the DR meter. Download the attached files.

Get the free encoder pack:
https://www.foobar2000.org/encoderpack

Now encode white.wav (DR: -2) to some lossy formats with different settings. Here are the results:

DR Peak RMS Duration Track
--------------------------------------------------------------------------------
DR6 -2.91 dB -9.49 dB 0:10 ?-vorbis q5
DR0 -10.72 dB -9.07 dB 0:10 ?-vorbis q10
DR6 -3.48 dB -9.49 dB 0:10 ?-opus vbr 128
DR5 -4.38 dB -9.50 dB 0:10 ?-opus vbr 256
--------------------------------------------------------------------------------

 

[bennetng]

I made a tool called BitSort, it can be used with foobar2000, SoX and other software. It counts the samples in audio files and shows statistics. For example, here is a classical music CD, bits with more than 1% of total number of samples are highlighted:

...and an electronic music album, not particularly loud for this genre, though. Samples in bit 0 are mostly leading and trailing silence, ignore them.

Reason:
https://help.foobar2000.org/troubleshooter/components/27209dfe26d9450f
For example, if you use this plugin to scan tracker formats and MIDI files, foobar will crash.

Unlike DR meter, BitSort can scan everything foobar2000 can decode, let me know if it crashes.

Download link (Microsoft OneDrive)
https://1drv.ms/u/s!AvzB71jO7t0-gYwfVGulbuFbVFZcmg?e=uN2p8Z

SHA256:
ee3a919fb267b9afec9494a0ae7f9b3ebd4fad97b8a03cc65245839ec2138d4f

There are two programs, BitSort and oldsCool in the archive. BitSort is actually a child project to help me debug oldsCool, a tool to convert some Cool Edit Pro/Adobe Audition specific .wav file formats to standard .wav formats. For more details, see ReadMe in the archive above.

 

[czt]

Obviously we don't have any problem with compression on drums, on vocals or for artistic purposes. But we don't want destroyed sound because of master bus compression just for the sake of "mastering for loudness" (war). And it is absolutely pointless in this days when streaming services is increasingly establishing loudness normalization.

-10.5 dB RMS (DR6) vs. -22 dB RMS (DR17):

https://forums.stevehoffman.tv/thre...heck-dynamic-range.326577/page-2#post-9446660