R128GAIN: An EBU R128 compliant loudness scanner Options

[jangk]

Have you tried explicitely setting mode, reference level and gate?

No, just using the standard (R128-2011)

Jean

[soundchekk]

Hi folks.

Just a quick question.

I'm wondering how ISO 226:2003 compares to all the algorithms supplied by r128gain??

Has anybody experience with that one??

THX
SC

[pbelkner]

alpha 6-2 is giving wrong results.

Many thanks for the hint. I've uploaded corrected source code and win32 binaries: http://sourceforge.net/projects/r128gain/files/r128gain/1.0/. Linux binaries will follow.

[jangk]

Many thanks for the hint.

You are welcome - thank you for your tremendeous work !

r128gain-1.0-alpha-6-3-win32-native performs perfectly.

Best regards
Jean

[pbelkner]

Linux binaries will follow.

The Linux binaries are available: http://sourceforge.net/projects/r128gain/files/r128gain/1.0/.

[unsword]

Thanks for the additon of the ATSC A/85 support and your continued work on this project.

Is there any difference between the following command line options ?

CODE

--a85 Run in ATSC A/85:2011 compliance mode.
--a85-2011 Run in ATSC A/85:2011 compliance mode.

[pbelkner]

Thanks for the additon of the ATSC A/85 support and your continued work on this project.

Is there any difference between the following command line options ?

CODE

--a85               Run in ATSC A/85:2011 compliance mode.
--a85-2011          Run in ATSC A/85:2011 compliance mode.

Currently there isn't any difference. The idea is that in case the standard evolves "--a85" should always refer to the latest version.

[soundchekk]

Did you manage to have a look at the two issues (script/coredump) I reported some posts back?

[soundchekk]

Did you manage to have a look at the two issues (script/coredump) I reported some posts back?

NO.

Just checked it out by myself.

[unsword]

Thanks for the additon of the ATSC A/85 support and your continued work on this project.

Is there any difference between the following command line options ?

CODE

--a85               Run in ATSC A/85:2011 compliance mode.
--a85-2011          Run in ATSC A/85:2011 compliance mode.

Currently there isn't any difference. The idea is that in case the standard evolves "--a85" should always refer to the latest version.

Thanks - I follow the logic

[soundchekk]

Hi.

--quiet

won't seem to work in latest alpha-6-3 cli-64.


Cheers

[soundchekk]

Hi there.

I'm preparing the big run now. On a couple of thousands songs I calculated more then 24 hours duration.

I tested the new Linux cli version ( as well as the old ones) on /tmp mounted into /dev/shm. And now on a HDD.

The HDD testrun is >40% slower, than my /dev/shm (RAMDISK). And even the duration of a conversion on ramdisk takes about 11s per track.


I'm wondering if r128gain uses a RAM buffer or if it stores the temporay files on disk.

If it buffers on HDD, things might get improved if I could select a "--temp=/tmp" option to use the RAMdisk as buffer.


What do you think?

Cheers

[Nologic]

Okay I just ran into a crash on WinXP SP3...probably asking for it to some degree.

Firstly I removed from .\r128gain-tools\:
avcodec-54.dll
avdevice-54.dll
avfilter-3.dll
avformat-54.dll
avutil-51.dll
ffmpeg.exe
postproc-52.dll
swresample-0.dll
swscale-2.dll

I then set FFMpeg's path via the command line:
r128gain.exe --ffmpeg=..\FFMpeg\ ...(other commands)...
or
r128gain.exe --ffmpeg="W:\My Projects\Video Gain\BIN\FFMpeg\" ...(other commands)...

Both pathing methods resulted in a crash...now if I don't remove the above files...everything is fine...but given that since those files are present in a normal shared FFMpeg release...shouldn't that make the ones present in R128Gain redundant...and thus unnecessary?

However in ether case...required or not...R128Gain should catch the error and simply bitch...not blow up.

Another thing to note...when I went:
r128gain.exe --help > ReadMe.txt

I got a blank file...and not one that contained all the command line help information.
r128gain.exe --help

Does display everything correctly in the console...but I can't seem to pipe it to a text file.

This post has been edited by Nologic: Oct 25 2012, 05:32

[Nologic]

Now I thought I read somewhere in the thread that R128Gain supported all formats that FFMpeg handles...It works fine with WAV & FLAC but when I try MP3 & AAC I get no love.

r128gain.exe --r128-2011 --true-peak --reference=-23 --in-place --tags=rg "C:\test\test.mp3"

Am I missing something?

[pbelkner]

Now I thought I read somewhere in the thread that R128Gain supported all formats that FFMpeg handles...It works fine with WAV & FLAC but when I try MP3 & AAC I get no love.

r128gain.exe --r128-2011 --true-peak --reference=-23 --in-place --tags=rg "C:\test\test.mp3"

Am I missing something?

You should update the FFmpeg DLLs from the sub-folder "r128gain-tools" with the corresponding from the latest Zeranoe build: http://ffmpeg.zeranoe.com/builds/win32/shared/.

[pbelkner]

I then set FFMpeg's path via the command line:
r128gain.exe --ffmpeg=..\FFMpeg\ ...(other commands)...
or
r128gain.exe --ffmpeg="W:\My Projects\Video Gain\BIN\FFMpeg\" ...(other commands)...

Both pathing methods resulted in a crash...now if I don't remove the above files...everything is fine...but given that since those files are present in a normal shared FFMpeg release...shouldn't that make the ones present in R128Gain redundant...and thus unnecessary?

I can't reproduce this one. Most likely you link to a binary incompatible FFmpeg. You should use the binaries from Zeranoe: http://ffmpeg.zeranoe.com/builds/win32/shared/

Another thing to note...when I went:
r128gain.exe --help > ReadMe.txt

I got a blank file...and not one that contained all the command line help information.
r128gain.exe --help

Does display everything correctly in the console...but I can't seem to pipe it to a text file.

For whatever reason R128GAIN writes the help text to stderr and not to stdout. Please use

CODE

r128gain.exe --help 2> ReadMe.txt

[Nologic]

Thanks for the quick reply.

I updated FFMpeg, both my independent install and the "r128gain-tools" which resulted in a working setup.

I was using a earlier build from zeranoe only a month or so old for my install.

So packaged R128Gain FFMpeg failed as noted prior...but once updated to the latest works fine.

However removing the files again as I noted in my prior post, and using --ffmpeg caused a crash yet again...so the --ffmpeg is mute since there is a dependency for having it in ".\r128gain-tools\".

Also --ffmpeg when using full path...if it has a trailing backslash it fails, but if it doesn't it works fine.
Works:
r128gain.exe --ffmpeg="W:\My Projects\Video Gain\BIN\FFMpeg" ...(other commands)..

Fails:
r128gain.exe --ffmpeg="W:\My Projects\Video Gain\BIN\FFMpeg\" ...(other commands)..

This issue isn't present when I performed relative pathing.
Works:
r128gain.exe --ffmpeg=..\FFMpeg\ ...(other commands)..
r128gain.exe --ffmpeg=..\FFMpeg ...(other commands)..

Thanks for all the help and effort on this project, it is appreciated.

[pbelkner]

Works:
r128gain.exe --ffmpeg="W:\My Projects\Video Gain\BIN\FFMpeg" ...(other commands)..

Fails:
r128gain.exe --ffmpeg="W:\My Projects\Video Gain\BIN\FFMpeg\" ...(other commands)..

This issue isn't present when I performed relative pathing.
Works:
r128gain.exe --ffmpeg=..\FFMpeg\ ...(other commands)..
r128gain.exe --ffmpeg=..\FFMpeg ...(other commands)..

Many thanks for the detailed report. The error seems to be special to XP (we had already another one of this kind). Unfortunately I don't have access to XP during the next few days.

[soundchekk]

Hi.

Since I can not run the r128gain in script mode ( the bug still exists), I'd like to make another proposal for the cli version.

I used to check the files if rpg tags existed prior ro conversion. This way I avoid conversions where not needed.

That's very convenient if you scan your entire music directory all the time . Or if the conversion stops (forced stop) and restarts somewhere in the middle of the process.

Basically what's needed is an option that allows to skip conversion in case RPG tags exist.


I really hope that the script bug gets fixed sooner or later. Otherwise usage of r128gain is very limited in cli mode. Above I could have easily implemented by myself.

THX.

[soundchekk]

Hi there.

I just had one of those Murphy's Law experiences.

A huge batch conversion which was running for hours, just caused a core dump .

It was finished with the analysis of one CD at that point in time.

When starting to write the data to disc, r128gain dumped with:

"11000 Floating Point exception (core dumped)."

What a mess.

Now I do have to run the whole stuff all over again, since the problem mentioned in my last post,
won't let my continue where the program caused the core dump.


A feature (option) that allows to skip analyse files with already written rpg(EBU)-tags would help a lot here.


Or even better - please try to fix my scripting issue. I can check it then by myself and run r128gain on a folder by folder basis.



THX

This post has been edited by soundchekk: Oct 26 2012, 13:32

[soundchekk]

What a disaster.

I just figured that r128gain messed up some of my files in that last directory, where the coredump took place.

I've got some dead bodies called:

X92jzl.flac
kuzTIz.flac
LLQuv7.flac



One hint: It's been Hirez 24/88.2 data and 28 files in a single directory or so. Maybe there's resource issue. I do have 8GB RAM in my machine.

[soundchekk]

Two more things on the CLI.

1. File permissions gets changed to 600 and ownership gets also changed by r128gain
It would be nice to protect file attributes or at least make sure that permissions are 666 after writing the new files.
2. While scanning the files a percentage counter runs. If you type "enter" that counter jumps randomly from one percentage to another ( far off) value.
Not sure if that has any impact on the analysis. It looks like problem on the first glance.



And one remark:

Just let me know if you're no longer interested in my feedback or the CLI version is put low on your priority list.
I noticed, that you obvisouly respond much quicker to other requests recently.
At least from my perspective - I'm encountering rather critical worst case issues - audio files are getting corrupted - with this app and I'm willing to support.
At least a - "I'll look into it (next week/month/year)" or "I can't reproduce it" would be appreciated.

THX

This post has been edited by soundchekk: Oct 27 2012, 10:01

[pbelkner]

I've uploaded a slightly enhanced version: http://sourceforge.net/projects/r128gain/files/r128gain/1.0/.

• Now it should be possible to run the Linux/Linux64 CLI versions as a deamon.
  

• Fails:
  r128gain.exe --ffmpeg="W:\My Projects\Video Gain\BIN\FFMpeg\" ...(other commands)..

  I was not able to reproduce this. But anyway, the new version removes a trailing slash.

• It would be nice to protect file attributes

  Now the Linux/Linux64 versions provide "--chown" and "--chmod" options for copying the respective attributes.
  Please note that overwriting the original files is not recommended. This is now emphasized by the help text.

Please note that FFmpeg's libavutil needs now to be version 52.

[soundchekk]

Info:

with Alpha-6-4 Linux-cli-64

--chown
--chmod
--progress=off
--quiet

don't work.

[Zao]

don't work.

What's the problem and how does it manifest? "Don't work" is a rather vague description