IPB

Welcome Guest ( Log In | Register )

 
Reply to this topicStart new topic
'splain to me how DR meters work, please
krabapple
post Mar 28 2014, 20:24
Post #1





Group: Members
Posts: 2330
Joined: 18-December 03
Member No.: 10538



Some (many?) audio file playback and editing tools offer 'dynamic range' (DR) analysis or metering..

I haven't used any of them, beyond simple post-hoc calculation of crest factor (peak amplitude - average RMS power) from file statistics that Audition dumps out , e.g.




Notice that it dumps stats for each channel.



What I'm wondering is, how do software DR meters come up with one 'DR' number for a stereo track -- i.e., one number for both channels?

And how would this work, if at all, on a 5.1 channel track , where there is typically large disparity in levels between channels like LFE versus front left?

This post has been edited by krabapple: Mar 28 2014, 20:29
Go to the top of the page
+Quote Post
skamp
post Mar 28 2014, 21:25
Post #2





Group: Developer
Posts: 1450
Joined: 4-May 04
From: France
Member No.: 13875



Sample output of the fb2k DR meter component (might be a clue in there):

CODE

foobar2000 1.2.9 / Dynamic Range Meter 1.1.1
log date: 2014-03-28 21:22:40

--------------------------------------------------------------------------------
Analyzed: Vangelis / Chariots of Fire
--------------------------------------------------------------------------------

DR Peak RMS Duration Track
--------------------------------------------------------------------------------
DR12 -6.56 dB -21.95 dB 3:33 01-Titles
DR12 -7.81 dB -23.72 dB 5:20 02-Five Circles
DR11 -9.25 dB -27.41 dB 3:14 03-Abraham's Theme
DR13 -0.44 dB -17.71 dB 4:19 04-Eric's Theme
DR14 -3.56 dB -23.93 dB 2:02 05-100 Metres
DR11 -1.40 dB -16.17 dB 2:48 06-Jerusalem
DR16 -0.67 dB -22.88 dB 20:42 07-Chariots of Fire
--------------------------------------------------------------------------------

Number of tracks: 7
Official DR value: DR13

Samplerate: 44100 Hz
Channels: 2
Bits per sample: 16
Bitrate: 626 kbps
Codec: FLAC
================================================================================



--------------------
See my profile for measurements, tools and recommendations.
Go to the top of the page
+Quote Post
AndyH-ha
post Mar 29 2014, 10:12
Post #3





Group: Members
Posts: 2210
Joined: 31-August 05
Member No.: 24222



There does not seem to be any fb2k DR meter on the page of foobar 2000 components http://www.foobar2000.org/components
Where do you find such a thing?
Go to the top of the page
+Quote Post
skamp
post Mar 29 2014, 10:19
Post #4





Group: Developer
Posts: 1450
Joined: 4-May 04
From: France
Member No.: 13875



QUOTE (AndyH-ha @ Mar 29 2014, 10:12) *
Where do you find such a thing?


Mirrored here and there.


--------------------
See my profile for measurements, tools and recommendations.
Go to the top of the page
+Quote Post
krabapple
post Mar 31 2014, 02:26
Post #5





Group: Members
Posts: 2330
Joined: 18-December 03
Member No.: 10538



QUOTE (skamp @ Mar 28 2014, 16:25) *
Sample output of the fb2k DR m


I've seen outputs; I'm curious as to how they are derived.
Go to the top of the page
+Quote Post
whitewidow
post Mar 31 2014, 07:18
Post #6





Group: Members
Posts: 11
Joined: 5-March 14
From: Austria
Member No.: 114824



"The DYNAMIC RANGE METER displays the inner dynamics of a recording in whole numbers or, in other words, the inner grade of compression. More precisely, this is the average cumulative difference between peak and loudness (RMS) over a specific period of time (e.g. duration of a song or album) and is a whole number value given in decibels (dB)." (http://www.tnt-audio.com/edcorner/february09.html)
Go to the top of the page
+Quote Post
krabapple
post Mar 31 2014, 23:25
Post #7





Group: Members
Posts: 2330
Joined: 18-December 03
Member No.: 10538




But how do they 'average' more than one channel?
Go to the top of the page
+Quote Post
whitewidow
post Apr 1 2014, 01:34
Post #8





Group: Members
Posts: 11
Joined: 5-March 14
From: Austria
Member No.: 114824



QUOTE (krabapple @ Apr 1 2014, 00:25) *
But how do they 'average' more than one channel?

In practice, the values don't differ much across channels. I think the creator was aware that DR is a crude measure. So maybe they're just taking the inf across channels (or the average across channels) even though that's not the rigorous thing to do.

This post has been edited by whitewidow: Apr 1 2014, 02:02
Go to the top of the page
+Quote Post
mjb2006
post Apr 1 2014, 05:31
Post #9





Group: Members
Posts: 829
Joined: 12-May 06
From: Colorado, USA
Member No.: 30694



Looks to me like it just gets separate DR values for each channel, then averages them.
Go to the top of the page
+Quote Post
user1
post Apr 8 2014, 05:45
Post #10





Group: Members
Posts: 39
Joined: 18-January 14
Member No.: 114096



Isn't (shouldn't) "DR" be standardized now (finally) to somehow jibe with the new international standards: E.g., LUFS, R128, ITU BS 1770, ATSC A/85 ...

Some more info here:

http://productionadvice.co.uk/lufs-dbfs-rms/
Go to the top of the page
+Quote Post
krabapple
post Apr 8 2014, 18:40
Post #11





Group: Members
Posts: 2330
Joined: 18-December 03
Member No.: 10538



QUOTE (whitewidow @ Mar 31 2014, 20:34) *
QUOTE (krabapple @ Apr 1 2014, 00:25) *
But how do they 'average' more than one channel?

In practice, the values don't differ much across channels.



They very much do, in 5.1 mixes. So I'm skeptical of a single 'DR' value for a multichannel mix.
Go to the top of the page
+Quote Post
markanini
post Apr 8 2014, 19:07
Post #12





Group: Members
Posts: 551
Joined: 22-December 03
From: Malmö, Sweden
Member No.: 10615



QUOTE (user1 @ Apr 8 2014, 06:45) *
Isn't (shouldn't) "DR" be standardized now (finally) to somehow jibe with the new international standards: E.g., LUFS, R128, ITU BS 1770, ATSC A/85 ...

Some more info here:

http://productionadvice.co.uk/lufs-dbfs-rms/


Bob Katz, Ian Shepherd and TC Electronics have discussed 'Peak-to-loudness ratio'. Googling reveals some interesting articles: http://www.google.com/search?q="peak+...ess+ratio"

This post has been edited by markanini: Apr 8 2014, 19:14
Go to the top of the page
+Quote Post
AliceWonder
post Apr 9 2014, 14:32
Post #13





Group: Members
Posts: 119
Joined: 13-July 12
From: California
Member No.: 101393



I don't put a lot of stock in DR.

There is a video somewhere on YouTube posted by the engineer who mastered the album showing the CD with a DR value of 8 while a needledrop from vinyl had a DR of 12.

Both were from the same identical master.

The differences when examined with a wav editor - in places where the master limited the peak, accurately reflected in the CD, the vinyl had some slightly different wave forms that he speculated was probably the result of the vinyl medium itself, the cartridge, tonearm, etc. in the analog chain between the vinyl and the ADC that recorded the needledrop to digital.

But the master was the same.

When listening to the portions he played, it certainly didn't sound like the vinyl had more dynamic range - it sounded the same.

Anyway point being for me, if identical master has DR of 12 just because there was a vinyle stage between the master and the tool used to calculate DR vs a DR 8 of the CD, then DR is not an accurate way of determining the actual dynamic range. A significantly higher number does not mean it actually has more range, so the value of DR is lost on me.

I expect a recording with low DR value does not have much dynamic range but the inverse is not true, higher DR value does not mean it has good dynamic range. The calculation method is flawed.

I gather it looks at the difference between peak volume and RMS volume, which sounds to be a reasonable method, but that YouTube demonstration shows that it is not actually very revealing and the value can vary wildly without having meaningful difference in the actual dynamic range.

I'll see if I can find the link.
Go to the top of the page
+Quote Post
AliceWonder
post Apr 9 2014, 15:09
Post #14





Group: Members
Posts: 119
Joined: 13-July 12
From: California
Member No.: 101393



http://www.youtube.com/watch?v=n-AE9dL5FG8

That's the video
Go to the top of the page
+Quote Post
markanini
post Apr 9 2014, 16:46
Post #15





Group: Members
Posts: 551
Joined: 22-December 03
From: Malmö, Sweden
Member No.: 10615



Just did a needle drop vs CD comparison(Allan Holdsworth - Secrets):

CD - Vinyl

Note the PRL values differs 0.1 dB.
Go to the top of the page
+Quote Post
markanini
post Apr 9 2014, 18:33
Post #16





Group: Members
Posts: 551
Joined: 22-December 03
From: Malmö, Sweden
Member No.: 10615



Did some more:

PRL CD | LP

Frank Zappa - Jazz From Hell
15.3 | 15.3

Jethro Tull - A
18.5 | 16.0

John McLaughlin - Electric Guitarist
16.5 | 14.6

Cream - Disraeli Gears
17.1 | 15.5

No inflated vinyl scores here it seems.
Go to the top of the page
+Quote Post
mjb2006
post Apr 9 2014, 23:19
Post #17





Group: Members
Posts: 829
Joined: 12-May 06
From: Colorado, USA
Member No.: 30694



QUOTE (AliceWonder @ Apr 9 2014, 08:09) *


This was discussed in the forums here; search for Ian Shepherd. Someone here noted that the one difference on vinyl, i.e. flat-topped peaks acquiring a downward slope, was due to a lowpass filter being applied during the vinyl mastering process. (Shepherd didn't seem to be aware that despite his delivery of just one master, there is still some further processing done for vinyl, e.g. to keep ultrasonic frequencies from overheating the lacquer cutting head.)

This rolloff of the highest frequencies, and the resulting change in the slope of the peaks, was all it took to get a higher DR value. You can also get a higher DR value by reducing the bass... anything you do to lower the peaks without lowering the average by an equal amount will, by definition, give you a lower DR value and shouldn't be surprising. Thus when using the same record, two needledrops made with different equipment can end up having different DR values, due to the different frequency response curves of the analog components.
Go to the top of the page
+Quote Post
markanini
post Apr 9 2014, 23:50
Post #18





Group: Members
Posts: 551
Joined: 22-December 03
From: Malmö, Sweden
Member No.: 10615



Ian Shepherd was eventually able to simulate the higher "DR" in TT meter on the digital master simply by narrowing the stereo image.
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 25th October 2014 - 09:03