Messages

Thanks for reporting this. I'm glad it works, I'll fix the "Unescaped left brace in regex is deprecated" warnings.
I'm not sure about the first error, about DBus, did it work before and does it work now or not ?

I just didn't have perl-net-dbus installed so it's natural this kind of error emerged, it's not critical thing (program runs fine without it) and I never needed it anyway so I didn't even notice it to this day. When installed there is no error and DBus control works.

I've updated perl from 5.20.2 to 5.22.0 yesterday and I got this trying to run gmusicbrowser (same result with 1.1.14 and git version):

Code Select Expand

Error loading gmusicbrowser_dbus.pm :
Base class package "Net::DBus::Object" is empty.
    (Perhaps you need to 'use' the module which defines that package first,
    or make that module available in @INC (@INC contains: /usr/bin/../share/gmusicbrowser /usr/lib/perl5/site_perl /usr/share/perl5/site_perl /usr/lib/perl5/vendor_perl /usr/share/perl5/vendor_perl /usr/lib/perl5/core_perl /usr/share/perl5/core_perl .).
at /usr/bin/../share/gmusicbrowser/gmusicbrowser_dbus.pm line 13.
BEGIN failed--compilation aborted at /usr/bin/../share/gmusicbrowser/gmusicbrowser_dbus.pm line 13.
Compilation failed in require at /usr/bin/gmusicbrowser line 410.
=> controlling gmusicbrowser through DBus won't be possible.

Unescaped left brace in regex is deprecated, passed through in regex; marked by <-- HERE in m/^sub { <-- HERE / at /usr/bin/../share/gmusicbrowser/gmusicbrowser_songs.pm line 1601.
Use of uninitialized value $thousandsep in substitution (s///) at /usr/bin/gmusicbrowser line 148.
/usr/bin/perl: symbol lookup error: /usr/lib/perl5/vendor_perl/auto/Glib/Object/Introspection/Introspection.so: undefined symbol: Perl_xs_apiversion_bootcheck
My system, 64bit Arch Linux, all standard vanilla perl packages.

Edit:
It seems perl-glib-object-introspection on my config wasn't rebuild properly when updating. Did it manually once more and gb started correctly and everything works fine now.

I can confirm this, it's fixed upstream now

it's perl 5.20 not perl 5.2 :)

haha my bad I am always getting caught with this  ;D

About the clipping prevention, if the problem is only that it doesn't work correctly, I'd rather fix it. I'll try to take a look and see if I see/hear a problem.

Personally I am using track gain tag as volume control for each track. Replaygain analysis just doesn't do the job right, especially with mix of old songs mastered in times when loudness wasn't top priority and new ones. When I got my whole collection through replaygain analysis volume was all over the place, much worse than before the whole process. That's why I got global volume on 60 and if some track is too quiet I just adjust it's track gain tag (entering manually also peak on advanced tags seems too tiresome) . Problem with clipping prevention is that it works the same regardless of global volume

example:
with clipping prevention
At 60% volume program is parsing
mplayer -volume 21.6
and with 1dB track gain tag (no peak selected so it's probably 1)
mplayer -volume 21.6
At 100% volume and replaygain analysis (4.26 dB gain, 0.73 peak)
mplayer -volume 136.85

without clipping prevention:
At 60% volume program is parsing
mplayer -volume 21.6
and with 1dB track gain tag
mplayer -volume 24.24
How can be clipping here? With 136.85 no clipping and with 24.24 there is?

I just want to have more control over volume for each track. I know I am doing it wrong (without lowering the peak) but I don't see any other way to easily get the right volume per track.

After upgrading perl to version 5.2 I get

Invalid version format (non-numeric data) at /usr/lib/perl5/core_perl/Encode.pm line 12.
BEGIN failed--compilation aborted at /usr/lib/perl5/core_perl/Encode.pm line 12.
Compilation failed in require at /usr/bin/gmusicbrowser line 26.
BEGIN failed--compilation aborted at /usr/bin/gmusicbrowser line 26.

as expected running with LC_ALL=C gmusicbrowser solves the problem but also killing localization support

Oh and can you make clipping prevention as option? I have a feeling it doesn't work as intended with mplayer backend and global volume other than 100. Whole replaygain analysis for mplayer seems a bit off or rather mplayer volume calculation (at least for my ears). I must change that line whenever I am updating gmusicbrowser and as I am using git it happens quite a lot (just call me lazy  ;D).

Thanks for fast reply, support from your side is always the best ;D

About the division by zero error, thanks, it should be fixed in git. It was a problem with your locale using a comma as decimal separator.

Yeah, the problem is gone now.

Some solutions :
1) ignore clipping prevention if there is no peak value, or use a higher default peak
2) make clipping prevention an option (with the additional problem that the gstreamer plugin doesn't offer this option, I'm not sure what its hard-limiter option do)
3) allow editing of peak

From solutions you've mentioned I would pick 1 (except I would completely ignore scale factor limitation) and 3. Why? Because if you use replaygain analysis it will stick to the standard one way or another (no clipping possible) and if you manually change replaygain values it means you don't like the effect of that analysis so why stick to it's boundaries. For global volume lower than 100 (in linear scale of mplayer)  there is no possibility of clipping so this is in some way unnatural. In my previous post volume stayed at 65.8503 where it shouldn't be any problem if it would rise to 100. For the third solution, we can edit peak with Advanced Tag Editing but making replaygain_track/album_peak editable in song properties dialog would be helpful, even if it's only for testing purposes.

I have commented the line you've suggested and the effect is what I was looking for. ;D

gain=0 peak=1 => scale factor=min(1,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 65.8503

gain=2 peak=1 => scale factor=min(1,25892541179417,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 82.9006160442694

gain=2 peak=2 => scale factor=min(1,25892541179417,0,5)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 82.9006160442694

with clipping prevention

gain=2 peak=1 => scale factor=min(1,25892541179417,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 65.8503

gain=2 peak=0,5 => scale factor=min(1,25892541179417,2)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 82.9006160442694

gain=2 peak=2 => scale factor=min(1,25892541179417,0,5)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 32.92515

It seems to work well, except that in any case above (without clipping prevention) there wasn't any sign of clipping at all. Replaygain analysis really fails with live music and amateur productions.


I found another bug:
1. Scan track with replaygain analisis
2. In song properities click save (no changes made)
3. Replaygain values are reset
It's related to the problem below. It happens with and without debug option

What happens exactly with Replaygain analyze in debug mode ? (I'll test it later)

Some sort of problem with tag saving/encoding. Replaygain tags are reset (rewritten with nan) after this. I don't have this issue outside debug mode.

Code Select Expand


*** GPerl asked to invoke callback from a foreign thread; handing it over to the main loop
done for ID=3267
replaygain-track-peak : 1
replaygain-reference-level : 89
replaygain-track-gain : -8,44000000000001
Wide character in warn at /usr/bin/../share/gmusicbrowser/gmusicbrowser_songs.pm line 1841.
Songs::Set(3267 ARRAY(0x4993fc0) abortmsg Przerwij analiz? Replay Gain error_prefix B??d zapisu znaczników Replay Gain :
abortcb CODE(0x49943e0) callback_finish CODE(0x49941a0)) called from : GMB::GST_ReplayGain:/usr/bin/../share/gmusicbrowser/gmusicbrowser_gstreamer-0.10.pm:658
HasChanged Progress -> updating Gtk2::HBox=HASH(0x52619b8)
HasChanged Progress -> updating Gtk2::HBox=HASH(0x52619b8)
FileTag::Write(3267,[replaygain_track_gain -nan replaygain_reference_level 89],CODE(0x50ec920))
new : Tag::ID3v2 Tag::MP3=HASH(0x50497f0)
ID3v2.3.0 : 69911 bytes
TIT3 (55 bytes)
TIT2 (13 bytes)
TIT1 (77 bytes)
TSSE (62 bytes)
TPE1 (21 bytes)
TCON (9 bytes)
TCOM (49 bytes)
TOPE (49 bytes)
TALB (19 bytes)
TYER (6 bytes)
USLT (29 bytes)
APIC (17457 bytes)
GEOB (51539 bytes)
TXXX (31 bytes)
TXXX (44 bytes)
TXXX (25 bytes)
........padding
length estimation : 185.88 s
startaudio=69911 copybegin=69911 length(towritebefore)=69601
endaudio=7505111 copyend=7505111 length(towriteafter)=128
in place editing.
ID=3267 towrite : replaygain_track_gain -nan replaygain_reference_level 89
HasChanged Progress -> updating Gtk2::HBox=HASH(0x52619b8)
new : Tag::ID3v2 Tag::MP3=HASH(0x4fbd598)
ID3v2.3.0 : 69911 bytes
TIT3 (55 bytes)
TIT2 (13 bytes)
TIT1 (77 bytes)
TSSE (62 bytes)
TPE1 (21 bytes)
TCON (9 bytes)
TCOM (49 bytes)
TOPE (49 bytes)
TALB (19 bytes)
TYER (6 bytes)
USLT (29 bytes)
APIC (17457 bytes)
GEOB (51539 bytes)
TXXX (31 bytes)
TXXX (25 bytes)
........padding
length estimation : 185,88 s
Songs::Changed : IDs=3267 fields=replaygain_track_gain modif
SongsChanged ARRAY(0x50e82d8) ARRAY(0x50d0fd8)
HasChanged CurSong -> updating CODE(0x3c4e3a8) Gtk2::Label=HASH(0x5246060) Stars=HASH(0x510ffa0) Layout::Label=HASH(0x5248a98) Layout::Label=HASH(0x5248510) Layout::Label=HASH(0x524c9a8) Gtk2::Table=HASH(0x524da80) Layout::AAPicture=HASH(0x5249ef8) Layout::Label::Time=HASH(0x510f048) Layout::Bar=HASH(0x52508d0) GMB::Plugin::LYRICS=HASH(0x51103a8) Layout::PictureBrowser=HASH(0x510f5c8) Gtk2::Label=HASH(0x52501b0) Gtk2::Label=HASH(0x525bee0) GMB::Plugin::LYRICS=HASH(0x510ddf0) Gtk2::Label=HASH(0x525d3a8) Gtk2::Label=HASH(0x525ff88) Layout::Label=HASH(0x510efa0) Layout::Window=HASH(0x510f550) GMB::DBus::Object=HASH(0x4f1daa8)
HasChanged SelectedID_Global(84997720-2) -> updating
HasChanged SelectedID_Global(84997720-3) -> updating
HasChanged SelectedID_Global(84997720-0) -> updating Layout::NoteBook=HASH(0x510f658)
HasChanged SongsChanged -> updating CODE(0x2cb0be8) SongList=HASH(0x5245708) FilterPane=HASH(0x51103d8) FilterPane=HASH(0x3081858) SongList=HASH(0x191b4d8) SongList=HASH(0x5110348) SongList=HASH(0x52600d8) SongList=HASH(0x525cd18) FilterPane=HASH(0x5250f88)
makesub filter=
makesub filter=
makesub filter=
HasChanged RecentSongs -> updating CODE(0x3c4e1b0)
IdleLoop End


It seems that everything works well with
LC_ALL=C gmusicbrowser -debug

Code Select Expand

*** GPerl asked to invoke callback from a foreign thread; handing it over to the main loop
done for ID=3267
replaygain-track-peak : 1
replaygain-reference-level : 89
replaygain-track-gain : -8.44000000000001
Songs::Set(3267 ARRAY(0x3bd2b58) abortmsg Abort ReplayGain analysis error_prefix Error writing replaygain tags :
abortcb CODE(0x3bd2f78) callback_finish CODE(0x3bd2d38)) called from : GMB::GST_ReplayGain:/usr/bin/../share/gmusicbrowser/gmusicbrowser_gstreamer-0.10.pm:658
HasChanged Progress -> updating Gtk2::HBox=HASH(0x44a1138)
HasChanged Progress -> updating Gtk2::HBox=HASH(0x44a1138)
FileTag::Write(3267,[replaygain_track_gain -8.44000000000001 replaygain_reference_level 89],CODE(0x432b1f8))
new : Tag::ID3v2 Tag::MP3=HASH(0x3946c70)
ID3v2.3.0 : 69911 bytes
TIT3 (55 bytes)
TIT2 (13 bytes)
TIT1 (77 bytes)
TSSE (62 bytes)
TPE1 (21 bytes)
TCON (9 bytes)
TCOM (49 bytes)
TOPE (49 bytes)
TALB (19 bytes)
TYER (6 bytes)
USLT (29 bytes)
APIC (17457 bytes)
GEOB (51539 bytes)
TXXX (31 bytes)
TXXX (25 bytes)
........padding
length estimation : 185.88 s
startaudio=69911 copybegin=69911 length(towritebefore)=69655
endaudio=7505111 copyend=7505111 length(towriteafter)=128
in place editing.
ID=3267 towrite : replaygain_track_gain -8.44000000000001 replaygain_reference_level 89
HasChanged Progress -> updating Gtk2::HBox=HASH(0x44a1138)
new : Tag::ID3v2 Tag::MP3=HASH(0x41fe220)
ID3v2.3.0 : 69911 bytes
TIT3 (55 bytes)
TIT2 (13 bytes)
TIT1 (77 bytes)
TSSE (62 bytes)
TPE1 (21 bytes)
TCON (9 bytes)
TCOM (49 bytes)
TOPE (49 bytes)
TALB (19 bytes)
TYER (6 bytes)
USLT (29 bytes)
APIC (17457 bytes)
GEOB (51539 bytes)
TXXX (31 bytes)
TXXX (44 bytes)
TXXX (25 bytes)
........padding
length estimation : 185.88 s
Songs::Changed : IDs=3267 fields=replaygain_track_gain modif
SongsChanged ARRAY(0x4303510) ARRAY(0x421eab0)
HasChanged CurSong -> updating CODE(0x2e7b010) Gtk2::Label=HASH(0x44857e0) Stars=HASH(0x4482de0) Layout::Label=HASH(0x4488218) Layout::Label=HASH(0x4487c90) Layout::Label=HASH(0x448c128) Gtk2::Table=HASH(0x448d200) Layout::AAPicture=HASH(0x4489678) Layout::Label::Time=HASH(0x434f848) Layout::Bar=HASH(0x4490050) GMB::Plugin::LYRICS=HASH(0x44831e8) Layout::PictureBrowser=HASH(0x434fda0) Gtk2::Label=HASH(0x448f780) Gtk2::Label=HASH(0x449b660) GMB::Plugin::LYRICS=HASH(0x434e5f0) Gtk2::Label=HASH(0x449cb28) Gtk2::Label=HASH(0x449f708) Layout::Label=HASH(0x434f7a0) Layout::Window=HASH(0x434fd28) GMB::DBus::Object=HASH(0x415dfe8)
HasChanged SelectedID_Global(71836824-0) -> updating Layout::NoteBook=HASH(0x4482498)
HasChanged SelectedID_Global(71836824-3) -> updating
HasChanged SelectedID_Global(71836824-2) -> updating
HasChanged SongsChanged -> updating CODE(0x1ee3e70) SongList=HASH(0x4484e88) FilterPane=HASH(0x4483218) FilterPane=HASH(0x22b4550) SongList=HASH(0xb97540) SongList=HASH(0x4483188) SongList=HASH(0x449f858) SongList=HASH(0x449c780) FilterPane=HASH(0x4499378)
makesub filter=
makesub filter=
makesub filter=
HasChanged RecentSongs -> updating CODE(0x2e7ae18)
IdleLoop End


Ok, I've checked the latest git version and there are several problems with replaygain.

For 90/100 volume in gmb without track gain set, album mode 0ff, pre-amp 0, fallback-gain 0   I get
gain=0 peak=1 => scale factor=min(1,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 72.9

the same but with pre-amp 10
gain=10 peak=1 => scale factor=min(3,16227766016838,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 72.9

For 87/100 volume
gain=9,99999999999999 peak=1 => scale factor=min(3,16227766016838,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 65.8503

For 0,3 dB track gain (set with advanced tag edit) and pre-amp 0
gain=0 peak=1 => scale factor=min(1,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 65.8503

For 1 dB and pre-amp 0
gain=1 peak=1 => scale factor=min(1,12201845430196,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 65.8503

For 4 dB and pre-amp 10
gain=14 peak=1 => scale factor=min(5,01187233627272,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 65.8503

For 10 dB and pre-amp 10
gain=20 peak=1 => scale factor=min(9,99999999999999,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 65.8503

For 20 dB and pre-amp 0
gain=20 peak=1 => scale factor=min(10,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 65.8503

For -1dB and pre-amp 10
gain=8,99999999999999 peak=1 => scale factor=min(2,81838293126445,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 65.8503

For -10 dB and pre-amp 10
gain=-1,06581410364015e-14 peak=1 => scale factor=min(0,999999999999999,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 65.8502999999999

For -1 dB and pre-amp 0
gain=-1 peak=1 => scale factor=min(0,891250938133746,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 58.6891416513886

For -10 dB and pre-amp 0
gain=-10 peak=1 => scale factor=min(0,316227766016838,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 20.8236932605386

For -15 dB and pre-amp 0
gain=-15 peak=1 => scale factor=min(0,177827941003892,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 11.7100232634886

For -20 dB and pre-amp 0
gain=-20 peak=1 => scale factor=min(0,1,1)
/usr/bin/mplayer -nocache -slave -vo null -nolirc -hr-mp3-seek -msglevel all=1:statusline=5 -softvol -volume 6.58503

It seems everything above 0 gain is ignored. Changing pre-amp and fallback-gain doesn't change anything (gain changes but volume not). You can allow higher values of -volume than 100 but I think you should use mplayer -softvol -softvol-max <10.0-10000.0> instead. There is also mplayer -softvol -af volume=x (where x is dB gain) but this won't work with mplayer -softvol -volume which you use. There are errors in mplayer documentation (or maybe it's so outdated) because you can set volume in range <0-10000> and setting softvol-max higher than 100 gives you louder sound without touching volume despite of what's written there. From my experience volume is usable in range <0-150>, 150 is about +11 dB which makes sense, with higher values there will be too much distortions in most cases.
mplayer -softvol -volume 150
and
mplayer - softvol - volume 100 -softvol-max 150
seems the same for me




There are also tracks that won't play using replaygain because of division by zero error. Output for track gain 0.389999999999986 dB and 0.702288627624512 peak (values from replaygain analyze)

Argument "0,39" isn't numeric in addition (+) at /usr/bin/../share/gmusicbrowser/gmusicbrowser_mplayer.pm line 225, <$OUTPUTfh> line 1113.
Argument "0,702289" isn't numeric in division (/) at /usr/bin/../share/gmusicbrowser/gmusicbrowser_mplayer.pm line 228, <$OUTPUTfh> line 1113.
*** unhandled exception in callback:
***   Illegal division by zero at /usr/bin/../share/gmusicbrowser/gmusicbrowser_mplayer.pm line 228, <$OUTPUTfh> line 1113.
***  ignoring at /usr/bin/gmusicbrowser line 1412, <$OUTPUTfh> line 1113.

Replaygain analyze doesn't work in debug mode but it works great in normal mode without crashes with 1.260 glib-perl now ;D

Quite frankly, replaygain is the standard, and it seems to be working well, so adding another tag to do the same job doesn't look like a good idea.

About replaygain only working with gstreamer, I took a quick look at mplayer, and it looks like using the options "-af volume=x" were x is the replaygain value, would do most of the job, and would be very easy to add.

About replaygain not satisfying you, maybe what you need is a way to edit the gain manually ? That would definitely allow you to change the loudness of a track as you wish.

Fair enough. Adding replaygain support for mplayer backend and some option to manually edit the gain within gmb would be great and solve my problem completely.

Note that currently scanning replaygain from within gmb is extremely unstable, it will be stable when using libglib-perl 1.251 or later.

Yeah, I noticed. With 1.241 gmb crashes almost every time I try to scan a track.

I have several tracks that are too quiet and this feature will be really helpful. I know there is ReplayGain but it doesn't work with other backends than gstreamer and I hardly ever got satisfying results with it. My proposition is adding one more tag/option which stores volume values. Whenever track will be played the volume should be set according to tag (<0-100> range, where everything outside means default volume).

Some time ago mplayer fork has been released. It supports gapeless playback through -gapless-audio option and don't have issues with seeking and pause. It can replace mplayer but many people install it next to standard package as mplayer2 or mplayer locally. Possibility of choosing path to mplayer binary would be nice ;D.

main site of the project
http://www.mplayer2.org/

You were right, mplayer crashes with pause in latest builds. What's surprising, it only affects audio files, so I haven't noticed. On older version of mplayer everything works fine. Thank you for reply.

When I click pause, player goes to the beginning of the next song on my list. The bug doesn't occur with any other backend. I am using Beta version 1.1.7 and Mplayer SVN-r33253-4.5.2