Dynamic Range (DR) for Hires Releases?

[HomerJAU]

Anyone interested in getting a report of their DR from existing Foobar files for your album folders and using Windows can use Music Media Helper's latest version (released today) to scan sub-folders and get visual list or PDF:

There's a screenshot and info on this post:
https://www.quadraphonicquad.com/fo...ic-videos-mch-audio.22693/page-12#post-466133
NOTE: This will probably only work if you have your music stored in the format:
ROOT MUSIC FOLDER\ARTIST NAME\ALBUM NAME

I haven't tested it for other folder formats, it may still work...

————-
EDIT: I tested on a flat folder structure with folders like this:
ROOT MUSIC FOLDER\ANY ARTIST NAME AND ALBUM NAME\files

MMH will work as designed, make a list of all albums showing tags and DR from Foobar creates DR files
————-

 

[ssully]

Seriously if you refuse to downmix you're probably better off just getting the DR of front left + right. In two channel it's rare to nonexistent that the left channel is 'slammed ' and the right is not. So comparisons across 2 channel recordings is safe. In 5.1 on the other hand it's possible for front left and right to be slammed, but Center and/or LFE and/or surrounds, not to be. This variance between channels in the same track can make 5.1 DR comparisons misleading.

 

[The56Kid]

That would be an awesome feature to add to the Polls, Garry. Thanks!

 

[J. PUPSTER]

Just some early morning reading for me to reinforce the basic understanding of DR. If anyone knows of more detailed articles on the subject they like, throw it up please.

https://www.izotope.com/en/learn/5-mix-tips-you-need-to-improve-dynamic-range.html

 

[HomerJAU]

Here‘s more reading. Apart frome the main article check the Further Reading links at the end:

https://en.wikipedia.org/wiki/Loudness_war
This is a good read (pdf):

https://tech.ebu.ch/docs/techreview/trev_2010-Q3_loudness_Camerer.pdf

 

[jimfisheye]

I'll never suggest to NOT include something even slightly useful of course.
The dynamic range calculator can be skewed pretty easily though and doesn't always tell much. The LUFS loudness scale used now adds loudness over time to the equation and at least will tell apart volume war fare vs not. That's not dynamic range but gives you an idea of what to expect.

The bigger telltales are extreme brickwall limiting (with makeup gain boost) and the ice pick to the ear drum treble slams that have been in vogue for CD masters for some time. Sometimes the HD master is just a copy of the CD aimed and prepped master masquerading as HD. (Well, the file is still HD format. The masquerading is around the expected audiophile edition suggested by the HD format.) This is what we want to avoid purchasing. You pretty much have to give it a listen. There might be some kind of algorithm that could be created to 'look' at those factors. A strict meter can't really see that. A loudness meter like LUFS will give you a pretty good heads up though. -16 to -13 LUFS is considered proper. Streaming services are bumped up a smidge to -12 LUFS nowadays. Loud but still not completely terrible CDs are usually around -10 LUFS. Volume war territory is -9 to -6 LUFS. The treble eq boosts often done in CD edition mastering are the biggest offender hands down. Peak limiting up to 6db is still pretty transparent. You'll see brick wall from that visually but you have to go deeper than 6db before you start hearing mad artifacts. Some albums that might look bad from the brick wall look actually aren't harmed much by that. The shrill thing is intentional high end eq boost. You have to listen and critique for that.

 

[HomerJAU]

Thanks Jim. I’ll do a quick comparison of LUFS to the DR14 measurement (used in Foobar).

I‘m not if theres ‘conversion’ between them and I don’t think the DR14 method is documented now their website in no longer running.

My Music Media Helper can calculate LUFs with one of its 3rd party apps included but it can’t do DR14 calcs as I don’t know how its implemented.

DR14 values are what we’re all used to seeing due to it being the first readily available to measure loudness (DR), including its Foobar2000 add-on component and it’s use on the Loudness War database. I think we need to stick with it here.

 

[jimfisheye]

Yeah, I'm not in any way suggesting to not keep posting that! Just that I've seen those DR values range across the board while being meaningless in telling you if someone bludgeoned the poor recording with a 20db treble boost.

And of course any critique of the eq profile of sound is subjective!
Was it a mastering disaster or is the original mix to blame? The treble hyped CD business has kind of a telltale sound. I should say that as - boosting treble on a mix after the fact in mastering can have a telltale sound vs merely a bright mix. But it's still a subjective listening exercise.

I'll state defiantly though that the hyped sound I'm describing is often attributed to the format ("It's only 16 bit!" "It's only 44.1k") but it's actually 99.999999999% intentional "mastering" work. (The quotes speak to my opinion of that.) Take your favorite 24/96 recording and convert it to 44.1k and then reduce that to 16 bit. See if you can tell them apart. Even if you have a reference system and DACs that perform better at HD than SD (so... not reference quality DACS then!), you aren't suddenly hearing shrill volume war from the format reduction.

A good exercise is to line up different CD editions of an album in your favorite DAW. First normalize the volume between them to allow critique. (Turn the louder ones down to match the quieter ones, not the other way around!) Now A/B between them. You'll start to wonder if they grabbed a random meth addled street bum to master a few of them!

 

[HomerJAU]

I’ve added DR to the polls database. I’m definitely wanting some help to enter the values... its a little painful, I’ve never been into data entry. I did about 20 albums and thought “This is going to take forever” (700 releases).

I’m going to create a simple Excel spreadsheet, pre-populated with all the hires titles and circulate it to volunteers here to add DR values, once it has a fair bit of data, then I’ll import into the dB for inclusion in the new DR report.

“CALLING FOR VOLUNTEERS“
I‘ll be ready with the DR Spreadsheet in a couple of days. There’s no pressure on deadline, just enter what you’ve got that’s not already there.

DR Contributors (volunteers so far):
Me
Albertop
J. Pupster
Stupy

 

[J. PUPSTER]

I’ve added DR to the polls database. I’m definitely wanting some help to enter the values... its a little painful, I’ve never been into data entry. I did about 20 albums and thought “This is going to take forever” (700 releases).

I’m going to create a simple Excel spreadsheet, pre-populated with all the hires titles and circulate it to volunteers here to add DR values, once it has a fair bit of data, then I’ll import into the dB for inclusion in the new DR report.

“CALLING FOR VOLUNTEERS“
I‘ll be ready with the DR Spreadsheet in a couple of days. There’s no pressure on deadline, just enter what you’ve got that’s not already there.

DR Contributors (volunteers so far):
Me
Albertop

I'd be happy to input values - I'm retired

 

[Stupy]

Can do... generating the DR report in foobar is part of my standard process...

 

[tlake6659]

I really like when members post waveform pictures as well for all the channels. This can show how compressed the front channels can be and is often better than just a DR number.

 

[HomerJAU]

Work on adding a DR report is going well. However DR for some SACDs is suspect (as mentioned early in this and other QQ threads).

Empty Channels and DR: MOST SACDs are not effected by this issue. We (the volunteers) have entered the DR as we've measured it. The quad AF and DV SACDs are the problem ones (mostly). If the silent channel is truly silent its ignored in the DR analysis and you get 4 channel DR divided by 4 = correct average. The AF/DV quads have noise or something else in the 'silent' centre, so the DR 'sees' the disc as 5.0. Sometimes that low level centre must have a DR of 50 or something silly so the average DR becomes very high, over 19 or 20. With others the centre is not ignored and the average DR is divided by 5 (not 4) for a low value.

My plan is to add a new Remix option to MMH to recursively remove a center and add new one so we can re-measure DR for those odd ball SACDs. (adding a silent centre in MMH using Sox has no DR problem). That may fix the DV DRs which maybe low as album DR maybe divided by 5 instead of 4 (quad) and the AFs with abnormally high DR.

So first publication of the new DR report will show some abnormal DRs until we re measure (using above) and there'll be a few holes where we don't have releases to measure, or we haven't measured those we do have yet.

At some stage I'll start a new thread for users to post their own DRs from missing on the report.

 

[HomerJAU]

The MMH update that adds a new Refresh Silent Channels option is complete (minus sanity testing). I added support for remixing channel layouts for DSF files too which will make life easier for users with DSFs.

This new option allows users to remove ’troublesome’ silent channels from files originating from quad SACDs and replace them with true silents (recursively in batch mode). Once that’s done DR measurement is no longer compromised and we get a true average a DR reading from just the 4 quad ‘real’ channels.

It should be released tomorrow (AU time)

 

[Beefalo]

The MMH update that adds a new Refresh Silent Channels option is complete (minus sanity testing). I added support for remixing channel layouts for DSF files too which will make life easier for users with DSFs.

This new option allows users to remove ’troublesome’ silent channels from files originating from quad SACDs and replace them with true silents (recursively in batch mode). Once that’s done DR measurement is no longer compromised and we get a true average a DR reading from just the 4 quad ‘real’ channels.

It should be released tomorrow (AU time)

 

[HomerJAU]

I've uploaded a new MMH version that now supports channel remixing for DSF files.

The bad news is that removing/adding new silents in DSF still gives incorrect DR readings!!! There's still noise in the silents using DSD, maybe its Foobar's DSD decoding that wonky?

Don't use the new 'Refresh Silents' option for DSF as that still has DR problems (above). You can still Refresh Silents for FLAC conversions from SACDs to fix DR readings

But you can use MMH to remove any existing silent channels in DSF then measure the new Quad DSF file for correct DR.

There is an option not the Overwrite source files, nothing is changed in the source file with that option. When you load files always make sure they are real quad (silent C and LFE) before removing the channels!!

This version loads files faster as it now using multiple threads. I still need to do work to speed up batch remixing, I think I can do better than current attempt. But that not critical now.

So we (volunteers ) will redo DRs for quad SACDs using either remixed DSFs and/or FLAC conversions over next fear days, then I’ll publish the report to QQ.