Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Matt]

MC19 makes several important improvements to audio analysis and volume leveling:

• Adoption of the R128 industry standard to analyze the loudness and dynamic range of content
  
• Ability to analyze audio for video files, including surround sound
• Smarter Volume Leveling that automatically respects intentional between track levels when playing from an album
• Volume Leveling works together with Adaptive Volume's peak level normalization
  
• Peak level is reported in decibels, measured as an R128 compliant True Peak, and reported per channel

This is a big topic, so I'll try to revisit it and post more details about the changes and the motivation at a later date.

[6233638]

If you are using the oversampling method of calculating peak level as specified by the EBU, it might be worth changing this to "True Peak Level" (measured in dBTP) - you don't want to sell yourself short.

[rjm]

Will backward compatibility be maintained so we can use the previous audio analysis method for old files and the new method for new files?

If not I will likely turn this feature off because I do not want to reanalyze my entire library.

[6233638]

I think the problem is that mixing R128 and ReplayGain is not going to give good results.
ReplayGain is not nearly as good as R128 at leveling volume to the point that you don't have to touch the volume control any more.

[6233638]

While I will have to re-analyze my files once true peak level analysis has been added, I am wondering about how downmixing factors into this.
I have analyzed a number of videos now, from downloaded content (e.g. made for YouTube or other sites) to television shows and films.

I haven't had a proper look over all the data, but there seems to be a number of items where volume leveling indicates that a positive correction is required, but peak level is already -1dB or higher. (and that's likely to increase once true peak measurements are implemented)

For example, I have an episode of The Wire which has a volume level of 4.7 LU, but a peak level of -1.3 dB, so volume leveling plays it back at +1.3 dB.
Even if I use Internal Volume or Parametric EQ set to -6dB as a test, Audio Path still indicates that Volume Leveling is only making a +1.3 dB correction. (P.S. when using the default Noire theme, Audio Path is not accessible during video playback in fullscreen mode for some reason)


The EBU R128 papers mention that separate analysis should be performed after downmixing, but you seemed to think that it wouldn't matter.
Would analysis after downmixing change things, or would the end result be the same?


It's still going to be very useful to have volume leveling active for downloaded videos, because they are often poorly mastered and the files I have analyzed so far range from -15.9 LU all the way to +22.0 LU, but it seems that it may not be very effective for commercially produced content - at least not if the target is -23 LUFS and peak level should not exceed -1 dBTP.

I seem to recall something being mentioned about possibly using -31 LUFS for some content in one of the EBU papers, but at some point I do wonder if we're just throwing away too much dynamic range in order to level things out though. (and when talking about losing 5-bits, it makes me wish my DAC accepted a 32-bit input…)

[6233638]

A few more thoughts on analysis and volume leveling:

1. It would be great if analysis could be made to work smarter. Because analysis is now likely to be I/O bound (though maybe not as much, now that True Peak Level analysis is in) if there are files from multiple drives in the list, it should try to assign data from different drives to each CPU core.

2. Analysis should be prioritized by size, length, and bit-depth/sample rate, so that it gets through the maximum number of files in the minimum amount of time. I'm not sure which of these factors in the most when it comes to speed. (probably length, then sample rate?)

3. I think my system went to sleep last night when analyzing files. I had another process that was also keeping the system awake, and I think once that finished it went to sleep. (but that was only an hour or so before I went to use the computer)

4. With the multichannel Peak Level tags, it just returns [Varies] rather than an average for groups now.

5. Formatting in the Peak Level tag is still a bit difficult to read. Even without using a monospaced font, switching to monospaced characters (figure space, and figure dash) and changing the spacing a bit, in an attempt to keep each channel a similar length makes it a lot more legible:



The real solution would be to use a monospaced font, or align the channels some other way.
Note: I have found that a number of monospaced fonts do not support the figure space character, which is rather annoying. (but of course a regular space is now monospaced anyway)

6. Because analysis is now a lot more CPU intensive, there's a higher chance of it impacting playback, or other usage of the computer. If it's possible, moving this to a lower priority process would be really nice.

7. If the analysis window is currently open, selecting more tracks for analysis should add them to the bottom of the list.

8. Would it be possible to pause analysis, rather than being forced to stop it? Especially now that we are analyzing large files, there are times where I need to stop analysis temporarily because it's interfering with another operation on the computer. (e.g. I want to copy a file from a drive that is currently being analyzed)


9. When volume leveling is active, could internal volume or parametric EQ volume adjustments change the target level, rather than the post-gain level? If a file is not able to be fully normalized to prevent clipping, I'd like some way of increasing the headroom. I don't know that -23 LUFS is enough to properly level video on my system when I'm mixing stereo and downmixed 5.1/7.1 content.

[InflatableMouse]

A few more thoughts on analysis and volume leveling

Just wanted to give a big thanks for your contribution mate, its being appreciated!

[mojave]

I just tried to analyze a Blu-ray concert. The concert has two titles with a different part of the concert on each title. I have the main library entry and a Particle to handle the two titles. I first analyzed the main library entry. Then I analyzed the Particle. The Particle showed as already analyzed and the tags were filled even though it was a different title than the main library entry. I analyzed the Particle and the results were different so evidently both titles have been analyzed separately.

Also, the first time I ran Analyze Audio I had the stereo track selected. I changed it to DTS-HD and reran Analyze Audio. It considers it already analyzed and I had to uncheck "Skip analyzed files." I wonder if there is a way to keep analysis info for various audio tracks, or at least to reset the analysis state if a track has been switched.

Summary:  Particles and different audio tracks show as having been analyzed if the either the main library entry or other audio track has been analyzed.

[6233638]

I just tried to analyze a Blu-ray concert. The concert has two titles with a different part of the concert on each title. I have the main library entry and a Particle to handle the two titles. I first analyzed the main library entry. Then I analyzed the Particle. The Particle showed as already analyzed and the tags were filled even though it was a different title than the main library entry. I analyzed the Particle and the results were different so evidently both titles have been analyzed separately.

Also, the first time I ran Analyze Audio I had the stereo track selected. I changed it to DTS-HD and reran Analyze Audio. It considers it already analyzed and I had to uncheck "Skip analyzed files." I wonder if there is a way to keep analysis info for various audio tracks, or at least to reset the analysis state if a track has been switched.

Summary:  Particles and different audio tracks show as having been analyzed if the either the main library entry or other audio track has been analyzed.

I don't know that there's a good or simple solution for this other than to either use the "best" track, or previously selected track if the file has already been played - which I think Media Center is already doing.

It's not an issue I have run into, or anticipate running into, because I just rip the best track and let JRSS handle the downmixing.
The problem is that if you're ripping the whole disc, that's potentially a lot of additional audio tracks you have to scan, if you want it to analyze everything. I have some films which have 5-10 audio tracks on them.

[mojave]

Just like you mentioned with iso's, I just rip the whole movie with JRiver so I don't have to think about anything. Later I can go back with MakeMKV and cut out stuff I don't want. However, I recently realized that I hadn't set MakeMKV to use the dtsdecoderdll.dll file and was only decoding the DTS core.  By ripping the entire movie, there are no mistakes.

I wouldn't want it to analyze all the audio tracks either, but if I ever switch tracks because the default chosen track is the wrong one, I would want JRiver to reanalyze based on the current track. What if the Peak Level for the stereo track was -5.0 dB and the DTS-HD was +2.3 dB? If it had originally analyzed the stereo track before I switched tracks and won't reanalyze, there could be a problem.

It is probably no big deal, but just something to think about.

[6233638]

I recently realized that I hadn't set MakeMKV to use the dtsdecoderdll.dll file and was only decoding the DTS core.  By ripping the entire movie, there are no mistakes.

I think this only matters if you are also converting to FLAC, but it's certainly easier to just rip to ISO.

I wouldn't want it to analyze all the audio tracks either, but if I ever switch tracks because the default chosen track is the wrong one, I would want JRiver to reanalyze based on the current track. What if the Peak Level for the stereo track was -5.0 dB and the DTS-HD was +2.3 dB? If it had originally analyzed the stereo track before I switched tracks and won't reanalyze, there could be a problem.

Well I don't think it should continue to use the same analysis, but I don't know that it should also force re-analysis. (because surely you're either playing the film, or finding the right track to rip if you're changing audio tracks)

[mojave]

I think this only matters if you are also converting to FLAC, but it's certainly easier to just rip to ISO.

You are right. I originally converted to FLAC but haven't the past few years. I haven't used MakeMKV on a Blu-ray in a long time and forgot about that.

[jack wallstreet]

I am pretty uneducated about this, but I have been using mp3gain to set levels for a number of years (instead of just MC) because mostly I play music on mp3players that aren't MC.  I wanted the files modified for a standard player.  Will MC19 be able to modify the files so they can play normalized outside of MC?  I hope?

[6233638]

I am pretty uneducated about this, but I have been using mp3gain to set levels for a number of years (instead of just MC) because mostly I play music on mp3players that aren't MC.  I wanted the files modified for a standard player.  Will MC19 be able to modify the files so they can play normalized outside of MC?  I hope?

I think MP3 Gain is unique in that it will modify the gain of your MP3s in a non-destructive manner.

You're probably best to stick with MP3 Gain and use MC analysis when playing inside Media Center.

[Quixote]

I am pretty uneducated about this, but I have been using mp3gain to set levels for a number of years (instead of just MC) because mostly I play music on mp3players that aren't MC.  I wanted the files modified for a standard player.  Will MC19 be able to modify the files so they can play normalized outside of MC?  I hope?

I think MP3 Gain is unique in that it will modify the gain of your MP3s in a non-destructive manner.

You're probably best to stick with MP3 Gain and use MC analysis when playing inside Media Center.

Thanks for asking and answering this question as I was doing/wondering the exact same thing.

[BigCat]

I am also in the same boat as Jack. I do MP3Gain first (for playing files outside my computer), then analyze from within JRiver. I have long thought that JRiver should add an MP3Gain-like modification during audio analysis. It would save me a lot of effort.

Though I am pretty uneducated about this as well, it would be great if JRMC19 would do this.

[MarkCoutinho]

I am also in the same boat as Jack. I do MP3Gain first (for playing files outside my computer), then analyze from within JRiver. I have long thought that JRiver should add an MP3Gain-like modification during audio analysis. It would save me a lot of effort.

Though I am pretty uneducated about this as well, it would be great if JRMC19 would do this.

One more guy who does it just like Jack, BigCat etc. I always use MP3Gain first and then MC. Would be great if I could throw MP3Gain away and use just MC for this purpose. Remember, like Jack said: Will MC19 be able to modify the files so they can play normalized outside of MC?

[Fred1]

One more guy who does it just like Jack, BigCat etc. I always use MP3Gain first and then MC. Would be great if I could throw MP3Gain away and use just MC for this purpose. Remember, like Jack said: Will MC19 be able to modify the files so they can play normalized outside of MC?

+1

see http://yabb.jriver.com/interact/index.php?topic=81626.0

[contium]

Can I get some clarification on the Dynamic Range (R128) measurement? I'm trying to understand how this measurement correlates with the dynamic range of the track and the DR value obtained from the Dynamic Range Meter/Dynamic Range Database.

Fore example, Muse's "The Second Law" is a horribly compressed CD:

Track   DR   Dynamic Range (R128)
      
1   DR5   7.0 LU
2   DR5   7.4 LU
3   DR6   3.6 LU
4   DR7   11.8 LU
5   DR5   14.2 LU
6   DR5   12.4 LU
7   DR5   5.2 LU
8   DR6   11.3 LU
9   DR6   7.7 LU
10   DR6   8.4 LU
11   DR4   2.3 LU
12   DR6   8.1 LU
13   DR5   13.6 LU

And The Cure's "Kiss Me Kiss Me Kiss Me" which is nice and dynamic:

Track   DR   Dynamic Range (R128)

1   DR11   8.9 LU
2   DR12   2.7 LU
3   DR12   2.9 LU
4   DR13   5.2 LU
5   DR12   2.6 LU
6   DR13   3.4 LU
7   DR13   3.5 LU
8   DR12   3.0 LU
9   DR11   3.9 LU
10   DR14   3.0 LU
11   DR12   3.0 LU
12   DR12   6.4 LU
13   DR12   2.5 LU
14   DR12   3.5 LU
15   DR12   6.5 LU
16   DR12   3.1 LU
17   DR12   4.5 LU

The R128 is all over the place, tells me nothing about what I might hear and almost seems inverted. Lower LU means more dynamic range? The DR values are much more consistent with what I hear.

[6233638]

A higher LU value should indicate higher dynamic range.

Media Center should be following the R128 spec: http://tech.ebu.ch/docs/tech/tech3342.pdf
I've just analyzed the EBU test files, and it seems to be working correctly.

[faster]

I think  R128 has nothing to do with Dynamic Range. Why did you name it Dynamic Range?

[Matt]

I think  R128 has nothing to do with Dynamic Range. Why did you namend it Dynamic Range?

http://tech.ebu.ch/docs/tech/tech3342.pdf

[faster]

http://tech.ebu.ch/docs/tech/tech3342.pdf

http://tech.ebu.ch/docs/tech/tech3342.pdf --> Loudness Range should not be confused with other measures of dynamic range or crest factor

This is what I mean!

[mojave]

The R128 is all over the place, tells me nothing about what I might hear and almost seems inverted. Lower LU means more dynamic range? The DR values are much more consistent with what I hear.

The TT Dynamic Range values are a measure of the crest factor of the music. Crest factor is the difference between average and peak levels in a track. This is based just as much on music style as whether there is compression used during mastering.

Actual dynamic range is the difference between the loudest and quietest parts of a track. The R128 values are a measure of the actual dynamic range of a track.

You can see from the Dynamic Range (R128) numbers from your two examples that "The 2nd Law" actually has more difference in dynamics than "Kiss Me Kiss Me Kiss Me." However, "The 2nd Law's" average levels are closer to its peak output which result in a much lower TT DR number.

This is what I mean!

The R128 Loudness Range is a measure of the difference between loudest and quietest portions of an audio track and should not be confused with TT Dynamic Range Meter which measures crest factor or the use of "dynamic range" to refer to the signal to noise content on a recording. JRiver is using Dynamic Range (R128) here because it probably has more meaning to people than Loudness Range.

[contium]

Hmm...I guess I was hoping to get a meaningful TT dynamic range like measurement. I don't see where MC19 returns the average R128 level. I'm assuming that would be -23 LU -(Volume Level(R128))? It would seem *in general* that a more negative Volume Level (R128) would indicate a more compressed/clipped file as long as the peak values don't drop as well. "The 2nd Law" averages about -15 LU for the Volume Level (128) with Peak Level (128) +0.5 dB while "Kiss Me Kiss Me Kiss Me" averages about -7LU for the Volume Level (128) and -0.5 dB Peak Level (R128). That would give "Kiss Me Kiss Me Kiss Me" about 7 LU more dynamic range which would correlate with the DR Database? Please correct me if I'm going off the deep end.

I do like the new volume leveling system.

[Matt]

Hmm...I guess I was hoping to get a meaningful TT dynamic range like measurement.

Next build of MC19:
NEW: Added DR dynamic range analysis to the audio analyzer.

[contium]

Next build of MC19:
NEW: Added DR dynamic range analysis to the audio analyzer.

Woo hoo!

[justsomeguy]

Quick question, is audio analysis pretty much how it's going to be going forward? Are we going to need to reanalyze our library again after this point? I only ask because MC has been analyzing my library now for 3.5hrs and it's only about 1/6th of the way finished. So I'm looking at about 21hrs total. Would hate to have to do it again.

[Matt]

Quick question, is audio analysis pretty much how it's going to be going forward? Are we going to need to reanalyze our library again after this point? I only ask because MC has been analyzing my library now for 3.5hrs and it's only about 1/6th of the way finished. So I'm looking at about 21hrs total. Would hate to have to do it again.

Since DR is coming next build (see above), you might wait to analyze.

[Vocalpoint]

Next build of MC19:NEW: Added DR dynamic range analysis to the audio analyzer.

Awesome!

VP

[hulkss]

Next build of MC19:
NEW: Added DR dynamic range analysis to the audio analyzer.

Very Nice! You guys seem to get to nearly all my requests. Some quick, some not, but definitely always a great effort on your part.

[contium]

This made the upgrade completely worth it. Thanks!

[mwillems]

Since DR is coming next build (see above), you might wait to analyze.

Currently reanalyzing my whole library, I'm super excited to try all this out.

[jkrzok]

Does anyone foresee any changes to audio analysis that may require a future reanalysis?

I have some 300,000 media files in my library and would like to save CPU cycles.

[astromo]

Since DR is coming next build (see above), you might wait to analyze.

Currently reanalyzing my whole library, I'm super excited to try all this out.

Yep, I hear ya. I've got a separate machine set up just to do this one job. Audio only files didn't take too long. Analysing fine video's the "fun" job...   

[mwillems]

Currently reanalyzing my whole library, I'm super excited to try all this out.


Yep, I hear ya. I've got a separate machine set up just to do this one job. Audio only files didn't take too long. Analysing fine video's the "fun" job...   

Just finished up my re-analyzing last night, and I have to say, I'm pretty impressed so far.  I cued up a play doctor playlist and listened for over an hour without having to adjust the volume once, which was never really the case with the old volume leveling (it would get close, but I'd find myself adjusting now and then even so).  It's also neat to see all the new info about tracks.  All in all a really neat improvement.

[6233638]

Just finished up my re-analyzing last night, and I have to say, I'm pretty impressed so far.  I cued up a play doctor playlist and listened for over an hour without having to adjust the volume once, which was never really the case with the old volume leveling (it would get close, but I'd find myself adjusting now and then even so).  It's also neat to see all the new info about tracks.  All in all a really neat improvement.

I agree - it's working really well when mixing random stereo tracks together.

When you introduce multichannel tracks - especially if you are downmixing to stereo - I think it could still use some work.
And album-based leveling now uses the gain from the loudest track rather than the average gain, which breaks leveling and has me changing the volume between albums again.

[JustinM]

In the Past I never saw the value in this feature...  Now that I've been listening to the r128 version..  Its a real treat not always playing with volume. It works great !
Congrats JRiver. an thanks for the feature I thought I didn't want 

[mwillems]

I agree - it's working really well when mixing random stereo tracks together.

When you introduce multichannel tracks - especially if you are downmixing to stereo - I think it could still use some work.
And album-based leveling now uses the gain from the loudest track rather than the average gain, which breaks leveling and has me changing the volume between albums again.

I noticed the change in the way albums were handled this morning.  I guess the idea is that it's designed to automatically prevent clipping if you use it with adaptive volume?  Not sure why else it would change.

I think I agree with you though, while it works great for playlists, it does seem result in some pretty serious inter-album differences. I just tried a "torture test": switching from Slayer (loudest track requires -16.5 LU) to Berlioz (loudest track requires -2.3 LU) (a 14.2 LU spread in the volume leveling).  To get those sounding about "the same" I had to make about a 3.5 dB adjustment, after volume leveling, to get subjectively similar sound (total of 17.7 difference).  The averages for those albums would have been around -16 LU and +2.2 LU respectively (an 18.2 LU spread). 

In this admittedly very small sample, the average would have covered my perceived difference in volume within a half dB, while the new "pick the lowest" method is much farther off.  I'm sure it won't work out that way in every case, but I'd be curious to hear additional observations on the new album leveling method.

[6233638]

I noticed the change in the way albums were handled this morning.  I guess the idea is that it's designed to automatically prevent clipping if you use it with adaptive volume?  Not sure why else it would change.

Yes, I believe the change was made to avoid potentially clipping with album playback. But with analysis measuring the peak levels, it seems like it shouldn't be necessary.

I think I agree with you though, while it works great for playlists, it does seem result in some pretty serious inter-album differences. I just tried a "torture test": switching from Slayer (loudest track requires -16.5 LU) to Berlioz (loudest track requires -2.3 LU) (a 14.2 LU spread in the volume leveling).  To get those sounding about "the same" I had to make about a 3.5 dB adjustment, after volume leveling, to get subjectively similar sound (total of 17.7 difference).  The averages for those albums would have been around -16 LU and +2.2 LU respectively (an 18.2 LU spread).

In this admittedly very small sample, the average would have covered my perceived difference in volume within a half dB, while the new "pick the lowest" method is much farther off.  I'm sure it won't work out that way in every case, but I'd be curious to hear additional observations on the new album leveling method.

This has been my experience too - there's a noticeable difference between playlists of albums now that it's using the loudest track's gain rather than the average.

I did some analysis on my library a couple of weeks ago, and about a third of my albums had a 3dB or greater delta, and 10% had a 6dB or greater delta when using the loudest track compared to the average level.
That's not to say that using the average level will be perfect either, but perceptually it seemed to give much better results.


I'm not convinced that clipping is going to be a widespread issue, because most music seems to be fine when normalized to -23 LUFS.
I do think we need some way to control the amount of headroom that volume leveling has available though; 23dB is not enough for video leveling - it needs more like 30dB.
Linking the internal volume control to the target volume level, rather than applying volume after leveling has been performed seems like the neatest solution for this. (if you want -30 LUFS as a target, reduce the volume by 7dB)

[MusicBringer]

Hello folks, I am new to this volume levelling (R128) feature.
I am confused.
Is it available to me now - if so how and where do I find it.
How do I use it; for example...what settings.
Is it still in beta and therefore subject to change.
thanks,

[6233638]

Hello folks, I am new to this volume levelling (R128) feature.
I am confused.
Is it available to me now - if so how and where do I find it.
How do I use it; for example...what settings.
Is it still in beta and therefore subject to change.

1. You need to analyze (or re-analyze) your files in MC19. Right-click your files and select Library Tools > Analyze Audio. This might take a while.
2. Enable Volume Leveling in DSP Studio

There have been a number of changes thus far, so we don't really know if there will be any more. It seems like it's in a state where we aren't going to see any more changes, but you may potentially have to re-analyze your library by the time MC19 is out of beta.

[Vocalpoint]

I'm not convinced that clipping is going to be a widespread issue, because most music seems to be fine when normalized to -23 LUFS.

Do remember that -23 LUFS is really designed for "broadcast" and keeping things under control in that environment. In my research on EBU 128 - I have seen several documents and studies that are promoting a movement of somewhere around -15 LUFS - being a more reasonable ballpark for music.

When I get MC 19 into my environment - I will be able to test this further....but in several previous tests with different software - i found -23 LUFS on a typical "non brickwalled" rock album to be excessive (to me) in driving down the levels too far.

YMMV.

Cheers!

VP

[mwillems]

I'm not convinced that clipping is going to be a widespread issue, because most music seems to be fine when normalized to -23 LUFS.
I do think we need some way to control the amount of headroom that volume leveling has available though; 23dB is not enough for video leveling - it needs more like 30dB.
Linking the internal volume control to the target volume level, rather than applying volume after leveling has been performed seems like the neatest solution for this. (if you want -30 LUFS as a target, reduce the volume by 7dB)

I had a question about that point, if you wouldn't mind expanding on it a little. My understanding is that internal volume happens "first."  I say that because most of the DSP modules have an option for "process independently of internal volume" that undoes internal volume for them (which only makes sense if internal volume happens first). So what would the advantage be (from a clipping perspective) of linking internal volume to volume leveling?  

Maybe an adjustable target for volume leveling might be a more straightforward solution, although I can understand if the devs are reluctant to move off of the studio standard.  

[6233638]

Do remember that -23 LUFS is really designed for "broadcast" and keeping things under control in that environment. In my research on EBU 128 - I have seen several documents and studies that are promoting a movement of somewhere around -15 LUFS - being a more reasonable ballpark for music.

When I get MC 19 into my environment - I will be able to test this further....but in several previous tests with different software - i found -23 LUFS on a typical "non brickwalled" rock album to be excessive (to me) in driving down the levels too far.

"Broadcast" includes radio - in fact that's where R128 has had the most adoption so far, I believe. Looking through the analyzed files in my library, 60% (of 15,000 files) require more headroom than -15 LUFS provides.

I had a question about that point, if you wouldn't mind expanding on it a little. My understanding is that internal volume happens "first."  I say that because most of the DSP modules have an option for "process independently of internal volume" that undoes internal volume for them (which only makes sense if internal volume happens first). So what would the advantage be (from a clipping perspective) of linking internal volume to volume leveling?

I thought it was generally recommended to not use the "process independently of independent volume" option for anything that is processing the audio, and it's just for visualizers/analyzers.

The issue is that, because volume leveling happens first, if it's going to run into clipping, it does not level fully.

For example, say a track has a peak level of -1dBTP, but would require a correction of +6dB to be leveled properly.
Volume Leveling will play this back at 0dB, because it cannot raise the volume any higher than that, so it's now playing at -29 LUFS.
If you reduce the volume control by 6dB, the track is now playing 12 dB lower than the target level (-35 LUFS) because the volume adjustment happens after leveling.


If Media Center's internal volume control adjusted the target level, reducing the volume control would give you additional headroom:
With the volume control at 100%, the track will still play back at 0dB, which is 6dB quieter than the target level.
If you reduce volume to 88% (-6dB) and that adjusts the target to -29 LUFS, it will still play back at 0dB, but other tracks will be up to 6dB lower, because the adjustment gave you an extra 6dB of headroom.
If you reduced the volume control to 76% (-12dB; -35 LUFS) then that track would be played back at -6dB. (because it's supposed to be +6dB from the leveling target)

Maybe an adjustable target for volume leveling might be a more straightforward solution, although I can understand if the devs are reluctant to move off of the studio standard.

Well the reason I suggest that it be integrated with the volume control is that it adjusts the target level without introducing another control. I don't see why you would want to use a "normal" volume control when leveling is enabled.

[Vocalpoint]

"Broadcast" includes radio - in fact that's where R128 has had the most adoption so far,

Haven't heard any radio stations here (Canada) that sound like they are using it - but I guess I should have specifically said television when I said "broadcast". With all the new "loudness" rules in effect here now - TV is where I notice LOTS of leveling has now occurred.

Radio has too many compressors and other doodads to make R128 worth the effort. And I do not thinking anyone is complaining about ads or music on the radio being too loud but I know (since I have actually called to complain) that ads on TV were out of control. I expect R128 plays a big role in that transition since it has become law now.

Looking through the analyzed files in my library, 60% (of 15,000 files) require more headroom than -15 LUFS provides.

I do not understand this - Can you explain how headroom a factor in applying an LUFS value?

Cheers,

VP

[mwillems]

"Broadcast" includes radio - in fact that's where R128 has had the most adoption so far, I believe. Looking through the analyzed files in my library, 60% (of 15,000 files) require more headroom than -15 LUFS provides.
I thought it was generally recommended to not use the "process independently of independent volume" option for anything that is processing the audio, and it's just for visualizers/analyzers.

The issue is that, because volume leveling happens first, if it's going to run into clipping, it does not level fully.

I think you might have it backwards (but I'm open to correction).  You're right that it is recommended not to use "process independently of internal volume," but in the recommended state (off), internal volume is happening before DSP modules.  Turning "process independently" on undoes internal volume for that module, which means that internal volume would otherwise happen first.  That's why setting the internal volume lower gives you more headroom for DSP that involve boost, like EQ or convolution. Matt, I think, has confirmed that internal volume gives you more DSP headroom in the PEQ and convolution contexts.  Otherwise it would be impossible to use any DSP that involved boost without constant risk of engaging clip protection, right?

If Media Center's internal volume control adjusted the target level, reducing the volume control would give you additional headroom:
With the volume control at 100%, the track will still play back at 0dB, which is 6dB quieter than the target level.
If you reduce volume to 88% (-6dB) and that adjusts the target to -29 LUFS, it will still play back at 0dB, but other tracks will be up to 6dB lower, because the adjustment gave you an extra 6dB of headroom.
If you reduced the volume control to 76% (-12dB; -35 LUFS) then that track would be played back at -6dB. (because it's supposed to be +6dB from the leveling target)

My understanding is that internal volume already gives you additional headroom, at least with other DSP modules (but I believe for volume leveling too).  For example, I have +5 dB shelf that, if it were processed before internal volume, would cause clipping all the time.  So I set my maximum internal volume -6 dBFS, with the result that I've never observed clipping in the analyzer even with very loud material.  It's possible that PEQ and Volume leveling are different in this respect?  

Thanks for the additional explanation, I'm a little dense sometimes    I'd be curious if one of the devs could confirm one or the other.  

[6233638]

I do not understand this - Can you explain how headroom a factor in applying an LUFS value?

1. R128 Peak level cannot be higher than -1.0 dBTP to avoid inter-sample clipping. (this is to account for some variance in the inter-sample peak calculation)
2. -23 LUFS is the target volume for leveling.

Using a real world example, you might have a track which has a [Peak Level (R128)] of -0.8 dBTP, and a [Volume Level (R128)] of 4.0 LU.
This means that the track needs to be played back at +4dB to sound like it's the same volume as other tracks.

Because clipping protection does not allow the peak level to exceed -1.0 dBTP, the track actually gets played back at -0.2 dB.
So to level this track properly, you need an additional 4.2dB of headroom; the target for leveling needs to be reduced to -27.2 LUFS, rather than -23 LUFS, for it to play back at the same perceived level as other tracks - either that or you just accept that this track is 4.2dB quieter than everything else.

If the target level was changed from -23 LUFS to -15 LUFS, this particular track would now be 12.2 dB quieter than any other track which is able to be normalized without clipping.

I think you might have it backwards (but I'm open to correction).  You're right that it is recommended not to use "process independently of internal volume," but in the recommended state (off), internal volume is happening before DSP modules.

Volume Leveling seems to happen first. This is the track I mentioned above, which is supposed to play at +4.0 dB:

Turning "process independently" on undoes internal volume for that module, which means that internal volume would otherwise happen first.

It's not an option for Volume Leveling.


EDIT:
In my library, I actually have a track (classical music) that needs an additional 7dB of headroom to be leveled properly, which would require the target be reduced to -30 LUFS!
But overall, -23 LUFS seems like a good target, as only 1% of my library (151 tracks) require more headroom than it provides.

This expression will let you see how much headroom your tracks require. (it could probably be cleaned up - but it works)

Code: [Select]

Delimit(if(isempty([Peak Level (R128)]),,formatnumber(math(removecharacters(left([Peak Level (R128)],5),/ /+,0)+RemoveCharacters([Volume Level (R128)],/ LU,0)+1),1)),/ dB,)
So if you were thinking about changing the target level to -15 LUFS rather than the current -23 LUFS, you would count all tracks that are -8.0 dB or greater. (note: positive values sort to the end of the list for some reason)


While -23 LUFS seems like a good target for music, video seems like it will require at least -30 LUFS to level properly, and we really need some way to analyze downmixed audio if multichannel files are being played back in stereo, for proper leveling.

[mwillems]

Because clipping protection does not allow the peak level to exceed -1.0 dBTP, the track actually gets played back at -0.2 dB.
So to level this track properly, you need an additional 4.2dB of headroom; the target for leveling needs to be reduced to -27.2 LUFS, rather than -23 LUFS, for it to play back at the same perceived level as other tracks - either that or you just accept that this track is 4.2dB quieter than everything else.

If the target level was changed from -23 LUFS to -15 LUFS, this particular track would now be 12.2 dB quieter than any other track which is able to be normalized without clipping.
Volume Leveling seems to happen first. This is the track I mentioned above, which is supposed to play at +4.0 dB:

Oh that ties that up; you're right (I was just looking at DSP studio myself and noticed what you describe). Internal volume happens before almost everything else, but not volume leveling.   I have large number of tracks/albums that need positive volume leveling adjustments but have near 0 dBFS peaks, and it sounds like they're unlikely to get the correct adjustments unless internal volume is applied first.  The vast majority of my listening happens at around 20% internal volume, so I'd have all the headroom I could ever need under those circs.

[mojave]

Using a real world example, you might have a track which has a [Peak Level (R128)] of -0.8 dBTP, and a [Volume Level (R128)] of 4.0 LU.

Do you have very many real world examples in your library where the Peak Level (R128) doesn't have enough headroom to adjust for Volume Level (R128)? In my library of 6400 audio tracks I found just three and it only made a 1.6 dB or less difference. I also would never play back those tracks without the entire album so it wouldn't matter anyway. The one needing the most Volume Level (R128) measures 15.8 and it has a Peak Level (R128) of -15.1 dBTP. The actual adjustment is 14.1 dB.

Only 200 tracks of mine need a positive volume adjustment (.03%). Out of those 200, there are only 4 tracks that I would play in a mixed playlist.