Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > General > Audio encoding

Reply
 
Thread Tools Search this Thread Display Modes
Old 26th November 2014, 02:39   #1  |  Link
manolito
Registered User
 
manolito's Avatar
 
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,078
EBU R128 Loudness Normalization

For the last two days I have educated myself a little about the EBU R128 recommendation for audio normalization. An AVStoDVD user over at videohelp requested that this standard should be integrated into the software (which creates DVDs, nothing that would be broadcasted).

I do understand the concept, but I am not sure if this is something that should be used for DVDs (or Audio CDs or BluRays). I see that users have a need for a uniform loudness perception across different media, but is EBU R128 the solution?

The standard has been developped for (and by) broadcasting people. Even before this standard was adopted, broadcasters peak-normalized their content to -9 dBFS which is much lower than what you see on audio CDs or DVDs (usually around -2dB to -3 dB). The reason is that in broadcasting the content is much more diverse compared to CDs or DVDs, with an emphasis on spoken content.

The overall result when applying this EBU R128 standard is that broadcasts are around 3 dB quieter than they were with the old -9 dB standard. This EBU standard will eventually be adopted for online content, too. But does it make sense for Audio CDs, DVDs or BluRays?


Any comments welcome...


Cheers
manolito
manolito is offline   Reply With Quote
Old 26th November 2014, 08:43   #2  |  Link
smok3
brontosaurusrex
 
smok3's Avatar
 
Join Date: Oct 2001
Posts: 2,392
short one: It does make sense, It is a thing of discipline, It does directly fight the loudness war. It does directly fight bad mixing/mastering "ideas".

example scans: https://bash-o-saurus-rex.googlecode...rockTop250.txt

p.s. If you are for some reason worried about levels being low, its probably easy to add some correctional number (user defined) to the calculated value.
__________________
certain other member

Last edited by smok3; 26th November 2014 at 08:55.
smok3 is offline   Reply With Quote
Old 26th November 2014, 08:55   #3  |  Link
Sparktank
47.952fps@71.928Hz
 
Sparktank's Avatar
 
Join Date: Mar 2011
Posts: 940
It really depends on the source material.

I find film scores to have the opposite effect desired when using the EBU r128 for normalization.
There are parts of musical composition (namely classical and score music (the "classical" music written for fim and television)) that are intended to be played quietly, but not so quiet you have to adjust your volume. You can hear it perfectly well on the current volume. EBU tends to dramatically increase that level and make it blare louder than the rest of the music that is naturally loud.

It's a good option if you know when and where to use it.

As a "night-time" mode for film, it would be good if you don't want to wake the whole house up and still enjoy the dynamic depth of audio of whatever show/movie you're watching.
I'd add a heavily compressed audio track as a secondary track on the authored disc as an option for such playback situations.

But, I think I'd rather go with using SoX for that type of dynamic compression. You're also given more control over boosting certain channels, like the center channel where a majority of the dialogue will be.

EBU R128 would be good for quick and dirty applications, for introductory levels.
__________________
Win10 (x64) build 19041
NVIDIA GeForce GTX 1060 3GB (GP106) 3071MB/GDDR5 | (r435_95-4)
NTSC | DVD: R1 | BD: A
AMD Ryzen 5 2600 @3.4GHz (6c/12th, I'm on AVX2 now!)
Sparktank is offline   Reply With Quote
Old 26th November 2014, 09:37   #4  |  Link
smok3
brontosaurusrex
 
smok3's Avatar
 
Join Date: Oct 2001
Posts: 2,392
Quote:
EBU tends to dramatically increase that level and make it blare louder than the rest of the music that is naturally loud.
You are in for some serious reading time, since you have absosmurfly no idea what you are talking about.
__________________
certain other member
smok3 is offline   Reply With Quote
Old 26th November 2014, 09:58   #5  |  Link
Sparktank
47.952fps@71.928Hz
 
Sparktank's Avatar
 
Join Date: Mar 2011
Posts: 940
Quote:
Originally Posted by smok3 View Post
You are in for some serious reading time, since you have absosmurfly no idea what you are talking about.
I'm talking about personal experience on some film scores that sucked using EBU R128.

This should be taken over to HydrogenAudio. They smurfing love this smurf.

On that note, I'll just mark down everyone's time of the month on a personal calendar and back out from here.
So many people here are becoming more and more hostile like animals in captivity that get too old and senile.
__________________
Win10 (x64) build 19041
NVIDIA GeForce GTX 1060 3GB (GP106) 3071MB/GDDR5 | (r435_95-4)
NTSC | DVD: R1 | BD: A
AMD Ryzen 5 2600 @3.4GHz (6c/12th, I'm on AVX2 now!)

Last edited by Sparktank; 26th November 2014 at 10:05.
Sparktank is offline   Reply With Quote
Old 26th November 2014, 17:44   #6  |  Link
pandy
Registered User
 
Join Date: Mar 2006
Posts: 1,049
Quote:
Originally Posted by smok3 View Post
short one: It does make sense, It is a thing of discipline, It does directly fight the loudness war. It does directly fight bad mixing/mastering "ideas".
Can you explain how it can prevent loudness war?

Whole concept of EBU R128 is wrong as it is in bad place (level equalization shall be performed at headend not at media).
In fact whole idea behind R.128 is to create good environment for advertisements.

Quality loss related to EBU R.128 implementation is to high...
pandy is offline   Reply With Quote
Old 26th November 2014, 22:25   #7  |  Link
detmek
Registered User
 
Join Date: Aug 2009
Posts: 463
Quality loss??? I use EBU R128 in foobar to make my music equally loud and it acts as automatic volume knob. How much quality can you loose by using volume knob?

It may reduce dynamic range if your audio file is 16-bit and has more then 86-90dB of dynamic range (like to see one, movie track or song) but human ear usually can't hear more then 35-40dB of dynamic range. And every song I have does not have nowhere near that high dynamic range.

You may change audio in unwanted way if you use plugin that adjusts loudness level by not scanning whole file. That leads to reducing dynamic range and altering loudness of some parts it unwanted or unexpected ways. So, just scan whole file and calculate loudness for it.

BTW, I asked a few months ago if it would be possible to add a feature to video players or audio decoders to read ReplayGain tags and adjust loudness during playback, just like audio players, but I didn't get any answer. That way audio would stay untouched and user can choose to use or not to use ReplayGain during playback. So far I can scan and tag MKV and MP4 containers using foobar2000.

Last edited by detmek; 26th November 2014 at 22:29.
detmek is offline   Reply With Quote
Old 26th November 2014, 23:21   #8  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,336
Quote:
Originally Posted by Sparktank View Post
But, I think I'd rather go with using SoX for that type of dynamic compression. You're also given more control over boosting certain channels, like the center channel where a majority of the dialogue will be.
What.

EBU R128 is not dynamic compression, not at all, not even close.

All it defines is a reference volume level and an algorithm on how to measure the volume level of a song/movie/etc.
Its basically ReplayGain with a more thought-out implementation and industry adoption, plus multi-channel support for movies.

It does NOT change the quality of the audio.
It does NOT change the dynamic range (unless like mentioned your sample used literally the entire dynamic range of the digital domain, which would not be perceivable by anyone)

All EBU R128 does is calculate a *static* volume adjustment for every movie, show or music track, so that they all sound about the same volume. Volume differences within one title are not affected (ie. the dynamic range stays untouched).
Music players generally also have a "album" mode where the average of the entire album is used to preserve intentional volume differences between tracks on the same album (assuming they are all mastered to the same reference anyway)

I can only second a previous comment about reading and understanding what EBU R128 actually does and is meant to accomplish.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders
nevcairiel is offline   Reply With Quote
Old 27th November 2014, 01:07   #9  |  Link
manolito
Registered User
 
manolito's Avatar
 
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,078
Thanks nevcairiel for pointing out all the important facts about EBU R128, couldn't have said it any better (I was just about to write a reply to the folks who quite obviously don't bother to enter a discussion without having a clue what they are talking about... )

Whatever, the AVStoDVD user who originally requested this feature made it clear that he has no intention to bring this standard to DVDs. He wants his DVD compilations (different tracks with different audio levels) to be watchable without correcting the playback audio level, and that from his experience the EBU R128 method to determine the correction factor is much more accurate than the ReplayGain method. And instead of the standard broadcast level of -23 LUFS he prefers a level of -11 LUFS (which IMO is a little high and requires a hard limiter at the end of the chain).

I think that I will add this feature to AVStoDVD, and I will include a user definable reference level to override the -23 LUFS default level. Thanks to ffmpeg, SoX and R128Gain this can be implemented quite easily using only open source tools.



Cheers
manolito

Last edited by manolito; 27th November 2014 at 01:10.
manolito is offline   Reply With Quote
Old 27th November 2014, 21:00   #10  |  Link
pandy
Registered User
 
Join Date: Mar 2006
Posts: 1,049
Quote:
Originally Posted by detmek View Post
Quality loss??? I use EBU R128 in foobar to make my music equally loud and it acts as automatic volume knob. How much quality can you loose by using volume knob?

Plenty - read about re-quantization already quantized signal.
http://en.wikipedia.org/wiki/Quantiz..._processing%29

Quote:
Originally Posted by detmek View Post
It may reduce dynamic range if your audio file is 16-bit and has more then 86-90dB of dynamic range (like to see one, movie track or song) but human ear usually can't hear more then 35-40dB of dynamic range. And every song I have does not have nowhere near that high dynamic range.
Any adjustment of the already quantized signal introduce distortions.

Of course there is possible to reduce those distortions but it imply usage of the:dithering, noise shaping and if possible oversampling.

Quote:
Originally Posted by detmek View Post
You may change audio in unwanted way if you use plugin that adjusts loudness level by not scanning whole file. That leads to reducing dynamic range and altering loudness of some parts it unwanted or unexpected ways. So, just scan whole file and calculate loudness for it.
In headend there is not possible to scan whole file - all adjustments need to be applied in real time, if your sources are from various providers then decoding -> adjustment -> reencoding is required - this lead to unavoidable loss of quality.

EBU R128 IMHO should be applied at CPE stage where correct requantization processing can be applied.

Quote:
Originally Posted by nevcairiel View Post
What.

EBU R128 is not dynamic compression, not at all, not even close.
True

Quote:
Originally Posted by nevcairiel View Post
All it defines is a reference volume level and an algorithm on how to measure the volume level of a song/movie/etc.
Its basically ReplayGain with a more thought-out implementation and industry adoption, plus multi-channel support for movies.
Partially true, first EBU R128 is based on Dolby Technical Bulletin 11 and as such it extend Dolby approach to perceived loudness (Dolby use different loudness reference level than analog studios and as such other audio codecs but at the same time Dolby use metadata to control final loudness and it is less affected by requantization problem). Dolby problem was that MPEG or AAC was perceived as louder when compared with Dolby (bad for Dolby business).
As i support industry good practices and i understand need of standardization for loudness level and loudness equalization then i disagree with proposed by EBU R128 concept of signal path.
IMHO loudness level measurement should be implemented at the end device and should be controllable by customer (i should be able to turn it off in same way as i can ignore Dolby metadata) - practical implementations for CPE are usually not inline with EBU recommendations.

Second EBU R128 cover various aspects of audio reproduction - it describe not only levels and how to measure but also describe detailed implementation of the signal path - this directly affect way how audio signals are decoded and processed - at the end of this path there DAC (and analog audio output) and/or digital audio outputs (S/PDIF, HDMI).

Quote:
Originally Posted by nevcairiel View Post
It does NOT change the quality of the audio.
Nope - after passing/propagating for fully EBU R128 compliant signal path audio level is different - at analog audio usually SNR is lowered (THD increased), on digital domain (S/PDIF, HDMI) only requantized signal (usually PCM) is available.


Quote:
Originally Posted by nevcairiel View Post
It does NOT change the dynamic range (unless like mentioned your sample used literally the entire dynamic range of the digital domain, which would not be perceivable by anyone)
It equalize loudness level to match/fulfill target value.

Quote:
Originally Posted by nevcairiel View Post
All EBU R128 does is calculate a *static* volume adjustment for every movie, show or music track, so that they all sound about the same volume. Volume differences within one title are not affected (ie. the dynamic range stays untouched).
Music players generally also have a "album" mode where the average of the entire album is used to preserve intentional volume differences between tracks on the same album (assuming they are all mastered to the same reference anyway)

I can only second a previous comment about reading and understanding what EBU R128 actually does and is meant to accomplish.
EBU R128 is more than loudness level meter and loudness level value, this is complex system to manipulate audio signal in real time and as such it directly affect audio signal quality.

I recommend EBU – TECH 3344 "Practical guidelines for distribution systems in accordance with EBU R 128" to understand implication for EBU R128 in practice.

Last edited by pandy; 27th November 2014 at 21:58.
pandy is offline   Reply With Quote
Old 27th November 2014, 22:23   #11  |  Link
smok3
brontosaurusrex
 
smok3's Avatar
 
Join Date: Oct 2001
Posts: 2,392
Quote:
IMHO loudness level measurement should be implemented and end device and should be controllable by customer (i should be able to turn it off in same way as i can ignore Dolby metadata).
That would mean changing all of the consumer devices, but yes.

Quote:
EBU R128 is more than loudness level meter and loudness level value, this is complex system to manipulate audio signal in real time and as such it directly affect audio signal quality.
example?

what is a "cpe stage" ?

Quote:
Any adjustment of the already quantized signal introduce distortions.

Of course there is possible to reduce those distortions but it imply usage of the:dithering, noise shaping and if possible oversampling.
I don't remember the multi-generation tests from this domain, any links? At what generation it does become audible?
__________________
certain other member

Last edited by smok3; 27th November 2014 at 22:36.
smok3 is offline   Reply With Quote
Old 28th November 2014, 04:45   #12  |  Link
manolito
Registered User
 
manolito's Avatar
 
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,078
Interesting discussion, but I'd like to bring this whole thing back to the real world...

No need to complain about this standard, for almost all TV and radio stations in my country (Germany) this standard is here, like it or not. The public TV stations started to adopt it in 2012, the private stations followed, and today the vast majority of all broadcasters observe it. And this is not about to please the advertisers as pandy suspects, just the opposite is true.

More and more viewers and listeners complained that commercials were much louder than the rest, these complaints even reached the European Commission. As a result a task force consisting mainly of broadcasting engineers was put together, and they hammered out this standard. (BTW it is a recommendation, not a mandatory standard).

My point is if and how this can be useful for DVD creation. There is a legitimate need for users to have a uniform loudness level for the multiple tracks of a compilation DVD. The older approach to achieve this was ReplayGain, but now everybody agrees that the loudness detection method of EBU R128 gives much better results.

The standard broadcasting level of -23 LUFS probably does not make much sense for DVDs, but of course the reference level can be brought up to -18 LUFS, maybe even to -14 LUFS. This brings the overall loudness to a level comparable with current commercial DVDs.

The problem with these higher reference levels is that the result might run into clipping. For my plugin implementation I use SoX to apply the gain correction determined by ffmpeg. As I see it SoX has two methods to deal with clipping:

1. Use the "gain" effect with the -l parameter.
2. Use the "vol" effect and add a "limitergain" parameter.

Both are not well documented. Right now I tend to use the second method with a limitergain value of 0.025. But I need to do more listening tests...
(any SoX specialists out there? )

Anyways, the concept works well, my favorite right now is to first apply DynamicAudioNormalizer (default settings) followed by EBU R128 loudness normalizing to -14 LUFS.


Cheers
manolito
manolito is offline   Reply With Quote
Old 29th November 2014, 10:50   #13  |  Link
hello_hello
Registered User
 
Join Date: Mar 2011
Posts: 4,821
My 2 cents.....

I've been using ReplayGain for years. I wouldn't leave home without it. Most of the MP3s on my MP3 player are ReplayGained courtesy of MP3Gain, although on occasion I've used foobar2000's ReplayGain scanner and MP3 volume adjusting capabilities instead. These days the foobar2000 ReplayGain scanner uses the EBU R128 standard instead of the original ReplayGain algorithm. I haven't compared them a lot but for standard audio (CD tracks) they don't seem to differ all that much (maybe 1dB or so).

Several times I've extracted the audio from a bunch of episodes of a TV show (from DVD) and run ReplayGain on them. ReplayGain tends to see them as having the same average volume, give or take a dB or so. If EBU R128 can do a better job for those making their own DVD compilations etc, that must be a good thing.

I've never quite understood why ReplayGain hasn't caught on when it comes to the audio accompanying video (in MKVs and MP4s etc), in respect to using ReplayGain tags and adjusting the volume on playback. It's fairly universally supported by software audio players, but then again, it's largely unsupported by hardware audio players.....
Even video conversion GUI's tend to completely ignore it as a rule and stick to "peak normalisation", which isn't always useful for keeping relative volumes the same. I use foobar2000 for most of my audio conversion so I scan and convert that way.

Quote:
Originally Posted by nevcairiel View Post
EBU R128 is not dynamic compression, not at all, not even close.
For the record, there's a foobar2000 "compressor" dsp that uses the EBU R128 algorithm/method for doing it's thing.
http://www.foobar2000.org/components/view/foo_r128norm
I don't really understand how it works exactly, but I assume that's what's being referred to. It does "compress".
I've tested it a little on some stereo soundtrack audio and it seems to work pretty well, but you've got no control over what it's doing.

If anyone cares..... there's a zip file at the end of this post containing some audio compression samples. The R128Norm DSP is one of them. I still prefer the WinAmp RockSteady DSP. These days if I was going to compress while re-encoding audio I'd probably do it with foobar2000, the Winamp DSP Bridge and RockSteady, or something similar.....
I use ffdshow and RockSteady to compress the audio on playback all the time (I downmix to stereo first).

Quote:
Originally Posted by manolito View Post
The problem with these higher reference levels is that the result might run into clipping. For my plugin implementation I use SoX to apply the gain correction determined by ffmpeg.
Are you still referring to using EBU R128 for running a ReplayGain scan? Wouldn't it work the same way, with the same option to adjust the volume according to the scan results and to adjust it further if need be to prevent clipping?
If so, couldn't you scan a bunch of audio files destined for a DVD compilation using "AlbumGain", then if any need to have their volume reduced further to prevent clipping, they'd all be reduced by the same amount to keep the relative volumes the same. That sort of thing......

Anyone know how the R128 levels of -23 LUFS and -18 LUFS etc translate into the old ReplayGain target volume of 83dB, or 89dB as is now the standard for music? I haven't got my head around that one yet.
hello_hello is offline   Reply With Quote
Old 29th November 2014, 23:31   #14  |  Link
nevcairiel
Registered Developer
 
Join Date: Mar 2010
Location: Hamburg/Germany
Posts: 10,336
The old ReplayGain reference would be around -18 LUFS (which is -18 dBFS), 5dB louder than the -23 LUFS EBU R128 reference.
And its a good thing that its lower too, since RG didn't always manage to create a uniform volume level since some tracks would clip otherwise, which should be a much rarer problem with the EBU R128 reference.

The biggest advantage of EBU R128 over RG is however proper and official support for multi-channel, which makes it actually usable for video.
__________________
LAV Filters - open source ffmpeg based media splitter and decoders
nevcairiel is offline   Reply With Quote
Old 30th November 2014, 03:14   #15  |  Link
manolito
Registered User
 
manolito's Avatar
 
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,078
@ hello_hello

I browsed through the discussion of this foobar2000 DSP, and it behaves quite differently compared to the original EBU R128 standard. While the EBU standard requires that the whole file is scanned first and then the gain is adjusted to reach the desired reference level, the foobar DSP works in 1-pass mode. It uses a lookahead buffer and adjusts the volume dynamically. This works like an Automatic Gain Control (sounds pretty similar to DynamicAudioNormalizer), and it does "compress" the overall dynamics of a stream.

In my AVStoDVD plugin you gan get this behavior by applying DynamicAudioNormalizer first and use EBU R128 Loudness Normalization as post-processing.


Like nevcairiel I also remember reading in one of the many articles I went over that to get a loudness matching ReplayGain the reference level should be around -18 LUFS. For my tests I used Video DVDs mostly, and using EBU R128 on them with -18 LUFS reduced the loudness a bit compared to the original.

If you want to make your own tests on material you already treated with ReplayGain, you can use current versions of FFmpeg to scan the files. Use this command line:
Quote:
ffmpeg.exe -nostats -i %1 -filter_complex ebur128 -f null -

Cheers
manolito
manolito is offline   Reply With Quote
Old 30th November 2014, 08:33   #16  |  Link
xooyoozoo
Registered User
 
Join Date: Dec 2012
Posts: 197
Quote:
Originally Posted by manolito View Post
If you want to make your own tests on material you already treated with ReplayGain, you can use current versions of FFmpeg to scan the files. Use this command line:
The filter works on a single file basis, but I'd like to obtain an album-gain value. Is there a standard, "correct" way of doing so?

I'm thinking of concatenating an aggregate macro file to run through ffmpeg, but it'd be nice to save an extra processing step if that value can be trivially calculated using per-track values.
xooyoozoo is offline   Reply With Quote
Old 30th November 2014, 09:19   #17  |  Link
hello_hello
Registered User
 
Join Date: Mar 2011
Posts: 4,821
Quote:
Originally Posted by manolito View Post
If you want to make your own tests on material you already treated with ReplayGain, you can use current versions of FFmpeg to scan the files. Use this command line:
I'll have a bit of a play soonish. Probably tomorrow.

I assume the EBU R128 "compressor" DSP uses the EBU R128 algorithm for determining volume, but it works with tiny sections at a time and adjusts the level accordingly. I have no doubt in that respect it works in a similar manner to DynamicAudioNormalizer.

The EBU R128 "compressor" DSP has nothing to do with foobar2000's normal ReplayGain scanning, which scans the whole file with the EBU R128 method and saves the results in tags etc, just as ReplayGain has always done.

I don't know if you listened to the samples I linked to but even though the EBU R128 DSP did a pretty good job of compressing the audio, it doesn't always compress short loud bits well. The sample contains a section of mostly low background noise, then a bunch of gunshots. The EBU R128 DSP actually amplified one of those gunshots a tad compared to the original audio.

Mind you it's hard to compare compression. I did it by running ReplayGain on each sample (and the original audio) and adjusted each accordingly. So they all have the same ReplayGain target volume after compression.

The one advantage the EBU R128 compressor DSP has over the other compressors is it's output file will effectively have the default ReplayGain target volume applied as a result of it's compression. Well, my little test encodes did at least. I ran ReplayGain on them and the EBU R128 compressor samples didn't require any further volume adjustment. The others did.

I included the original audio with the samples if you'd care to try DynamicAudioNormalizer as a comparison. I had a play with it a while back and found it quite good. The output sounded very much like the output from the RockSteady DSP (my compression settings) after I fiddled with a few of DynamicAudioNormalizer's options to suite my taste.

Off the topic of compression and back to normal ReplayGain......

One of the things I liked about MP3Gain for applying ReplayGain to MP3s is it lets you easily adjust the target volume and it tells you whether there'll be clipping as you do. I often lowered the target volume to 86dB or 83dB for "soundtrack" audio as it tends to me more dynamic (I think 83dB is the SMPTE standard) but I'd sometimes lower it enough so none of the audio would have peaks causing clipping and then I'd apply TrackGain. Thinking about it, it's probably the same end result as using "AlbumGain" with a further volume reduction as required to prevent clipping. Either way a group of audio files are all adjusted to the same target volume, but the manually changing the target volume method let me feel in control.

Quote:
Originally Posted by nevcairiel View Post
The old ReplayGain reference would be around -18 LUFS (which is -18 dBFS), 5dB louder than the -23 LUFS EBU R128 reference.
And its a good thing that its lower too, since RG didn't always manage to create a uniform volume level since some tracks would clip otherwise, which should be a much rarer problem with the EBU R128 reference.
ReplayGain using EBU R128 for scanning hasn't changed anything in respect to the ReplayGain's target volume.
Originally, the ReplayGain target volume was (I think) 83dB, but lots of people complained about it being too quiet, so it was pretty quickly changed to 89dB and it's been 89dB ever since. Unfortunately though, 89dB doesn't provide enough headroom for "soundtrack" audio, which is why I often lowered it to 86dB or 83dB when applying ReplayGain with MP3Gain.

Anyway, I dragged out an MP3 which I know was ReplayGained with MP3Gain and I checked the tags.
Track Gain -0.48dB, Peak 0.384223
I then ran a ReplayGain scan on it with Foobar2000 which apparently now uses the EBU R128 method for scanning.
Track Gain -0.56dB, Peak 0.384235

So they're a little different (it might even be due to the decoding) but they're basically the same. So whatever the EBU R128 equivalent of ReplayGain's target volume of 89dB might be (around -18 LUFS), the ReplayGain target volume hasn't changed, which makes sense I guess. It'd be a little odd if all the MP3's I scanned yesterday are now 6dB too loud because the ReplayGain scanning method was changed. Foobar2000 still shows the ReplayGain target volume as being 89dB by default (buried in it's preferences).

I guess talking about ReplayGain here is "my bad" so I hope I haven't confused things. This thread's about "EBU R128 Loudness Normalization" and I referred to "EBU R128 Loudness Normalization" and "ReplayGain" as being one and the same, given EBU R128 is sometimes used for ReplayGain scanning these days, and they would be, except it appears the original EBU R128 target volume has been increased for ReplayGain so it's not actually EBU R128 anymore. Sign.......
In my defence though, manolito referred to changing EBU R128 reference levels earlier, which probably helped shift my brain into ReplayGain mode.

I assume ffmpeg's EBU R128 volume adjustment would result in -23 LUFS as it should be. Or can you tell ffmpeg to apply ReplayGain too? Or would you just specify the appropriate target volume for ReplayGain in the command line? Although if ffmpeg doesn't apply ReplayGain as such, then you probably would miss out on some of the fun stuff, such as adjusting a group of files together via AlbumGain or using the ReplayGain "track peak" results to further adjust the volume and prevent clipping. They're the ReplayGain options I mentioned in an earlier post, in case what I was talking about wasn't clear at the time.

I'm too tired. Time for bed.

Last edited by hello_hello; 1st December 2014 at 18:29.
hello_hello is offline   Reply With Quote
Old 30th November 2014, 19:47   #18  |  Link
pandy
Registered User
 
Join Date: Mar 2006
Posts: 1,049
Quote:
Originally Posted by smok3 View Post
That would mean changing all of the consumer devices, but yes.
yes but IMHo in real life this is only one solution that can work - side to this with modern DSP present in each HDTV decoder imlementing device similar to LM100 is trivial.

Quote:
Originally Posted by smok3 View Post
example?

what is a "cpe stage" ?
Look at EBU – TECH 3344 - CPE means any customer device.

Quote:
Originally Posted by smok3 View Post
I don't remember the multi-generation tests from this domain, any links? At what generation it does become audible?
Be more specific - i don't understand "multi-generation tests" term - sorry.

Audible distortions - talk with vinyl hipsters - from mathematical perspective this is obvious - divide integer by fraction.
Re-quantization without additional processing will produce distortions spectraly correlated with signal.

Btw: EBU-R128 is not a standard - this is recommendation, it may be used voluntary by various bodies and similar end results are expected - but one again this is purely voluntary.

Last edited by pandy; 30th November 2014 at 19:54.
pandy is offline   Reply With Quote
Old 1st December 2014, 08:35   #19  |  Link
manolito
Registered User
 
manolito's Avatar
 
Join Date: Sep 2003
Location: Berlin, Germany
Posts: 3,078
@ hello_hello

I think you are confusing ReplayGain with EBU R128 a little bit...

When you are talking abut the foobar2000 ReplayGain using the EBU R128 method for scanning, I believe this is the wrong term. If loudness scanning is done using EBU R128, then it is NOT ReplayGain any more. I would call it EBU R128 with a different reference level (e.g. -18 LUFS instead of the standard -23 LUFS).

Both ReplayGain and EBU R128 establish a method for loudness scanning, plus they establish a reference loudness level. These two things are independent of each other. You can use ReplayGain scanning and set a reference level different from the standard 89 dB, and you can use EBU R128 scanning and also set a different reference level.

For EBU R128 the standard reference level of -23 LUFS has been established for broadcasting, not for other applications like Audio CDs, MP3, DVD or Blu-Ray. Even before EBU R128 the standard broadcasting level has always been lower than for other media (-9 dB peak), because broadcasting has to cover a much wider spectrum of different source material than other media.

For my plugin which covers only DVD creation it makes total sense to use the EBU R128 scanning method, but employ a higher reference level like -18 LUFS. Most people seem to agree hat the EBU scanning method delivers more consistent results than the RepülayGain method, and it handles 6-ch audio which ReplayGain does not.

For FFmpeg the EBU R128 scanning is totally separate from the following loudness adjustment. You have to do the scanning pass, note the LUFS value, calculate the difference to the desired reference value, and then do a second pass for the loudness correction. The R128Gain software can do it all in one step, using FFmpeg for scanning and either SoX or FFmpeg for the loudness adjustment.


Cheers
manolito
manolito is offline   Reply With Quote
Old 1st December 2014, 15:18   #20  |  Link
smok3
brontosaurusrex
 
smok3's Avatar
 
Join Date: Oct 2001
Posts: 2,392
Quote:
The R128Gain software can do it all in one step
And how is that supposed to work? (time travel of some sort?)
__________________
certain other member
smok3 is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 08:40.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.