This is Woof! 14.5.0 (Apr 30, 2024)

speaking of soundfonts, I've recently tried the latest EDGE-Classic release and discovered that it has adopted "GMGSx.SF2" as its default soundfont

I never heard of it before, but it sounds pretty close to RLNDGM.SF2 which seems to be the standard for most doomworld's midi composers (and sounds better imho than TIMGM6MB).

On top of that, the author released it as Public Domain: I did not find any explicit license document, but the the .SF2 file is included in the free (as in beer) midi editor/player SynthFont1 (developed by the same dude who created the soundfont)

does the GPL2 let developers redistribute PD content with their programs?

@FEZ_GG Did you try a recent AppImage or the one from the release? I'm not an AppImage expert myself, but I know we have hard-coded the soundfont directories, not the soundfont itself. So, in theory, your approach should work. 

25 minutes ago, liPillON said:

does the GPL2 let developers redistribute PD content with their programs?

Yes. But I would still want to read an explicit license grant by the copyright holder. A mere "this guy is cool, he had released other free stuff, so this should be free enough too" won't work. ;)

I thought so..

I just wanted to know what the GPL would have let you do, before investigating on SynthFont's licensing..

So I downloaded the setup package and used innoextract to get the Readme and the License files.

the former says nothing about the SF2 file

the latter also covers the software package only:

I guess they're both a no-no

You may want to mail him and ask for an explicit license for the sf2 file?

I'll try. 

34 minutes ago, fabian said:

@FEZ_GG Did you try a recent AppImage or the one from the release? I'm not an AppImage expert myself, but I know we have hard-coded the soundfont directories, not the soundfont itself. So, in theory, your approach should work.

I used the AppImage from the Github release.  The MIDI Player options I'm getting are DEFAULT-GM.SF2, FLUIDR3_GM.SF2, TIMGM6MB.SF2 and DEFAULT-GM.SF3 for Fluidsynth, and the OPL3 emulation.  I put RLNDGM in the soundfonts folder, but it didn't add any other options.

1 hour ago, liPillON said:

speaking of soundfonts, I've recently tried the latest EDGE-Classic release and discovered that it has adopted "GMGSx.SF2" as its default soundfont

I never heard of it before, but it sounds pretty close to RLNDGM.SF2 which seems to be the standard for most doomworld's midi composers (and sounds better imho than TIMGM6MB).

On top of that, the author released it as Public Domain: I did not find any explicit license document, but the the .SF2 file is included in the free (as in beer) midi editor/player SynthFont1 (developed by the same dude who created the soundfont)

 

does the GPL2 let developers redistribute PD content with their programs?

 

The biggest problem with many of these "public domain" soundfonts is that they almost always borrow samples from gm.dls, the default Windows soundfont. I downloaded the SynthFont app and opened GMGSx.SF2 side by side with gm.dls and sure enough the samples are identical.

gm.dls on the left, GMGSx.SF2 on the right:

They are byte for byte the same except the author removed the Roland copyright:

54 minutes ago, ceski said:

 

The biggest problem with many of these "public domain" soundfonts is that they almost always borrow samples from gm.dls, the default Windows soundfont. I downloaded the SynthFont app and opened GMGSx.SF2 side by side with gm.dls and sure enough the samples are identical.

 

gm.dls on the left, GMGSx.SF2 on the right:

 

 

They are byte for byte the same except the author removed the Roland copyright:

 

I opened Polyphone and looked to see if the Roland copyright was present; if it is otherwise identical I will pull it from our repo. Searching for open source friendly soundfonts has been a headache.

3 minutes ago, dasho said:

 

I opened Polyphone and looked to see if the Roland copyright was present; if it is otherwise identical I will pull it from our repo. Searching for open source friendly soundfonts has been a headache.

I checked about a dozen samples at random and they're all the same. The author sometimes applied modulators or pitch shifted but the underlying sample was always Roland's. It really is a headache, agreed.

Sorry for taking this tangent further but I've often wondered about the way copyright works when it comes to taking sounds from hardware/software meant to make music. Obviously just taking gm.dls and stripping out the copyright text is a no-go. Taking recordings of all the samples from it and then making a soundfont from that is also probably not valid. But then recording a song using gm.dls doesn't give Roland any copywrite on your recording. And if you made your recording public domain, and then someone sampled it? I guess maybe intent matters to some extent, but IMO it gets complicated.

The point of all that being, it's probably very hard to make a truly and legally public-domain soundfont that matches an existing one.

It might be interesting to try to make a community-sourced soundfont, but whoever is organizing it would have to be diligent about making sure the sources are free. And it's unclear to me how close to first principles you would need to get for making the sounds, especially synthesized ones.

9 hours ago, ceski said:

they are byte for byte the same except the author removed the Roland copyright:

ouch! 

yeah I don't think I'll write to the guy, after all

12 hours ago, FEZ_GG said:

I used the AppImage from the Github release.  The MIDI Player options I'm getting are DEFAULT-GM.SF2, FLUIDR3_GM.SF2, TIMGM6MB.SF2 and DEFAULT-GM.SF3 for Fluidsynth, and the OPL3 emulation.  I put RLNDGM in the soundfonts folder, but it didn't add any other options.

You can set the directory to read from in the config file (~/.local/share/woof/Woof-12.0.0-Linux.cfg or something like that), look for "soundfont_dir" (or just directly set "soundfont_path" to the soundfont you want to use).

11 hours ago, FEZ_GG said:

I used the AppImage from the Github release. 

This won't work, unfortunately. You'll need to use a more recent snapshot. The release only finds the system-wide installed soundfonts. I added the ability to search for soundfonts relative to the executable directory just after the 12.0 release. 

12 hours ago, ceski said:

They are byte for byte the same except the author removed the Roland copyright

Worst case, indeed, thanks for checking!

11 hours ago, plums said:

Sorry for taking this tangent further but I've often wondered about the way copyright works when it comes to taking sounds from hardware/software meant to make music.

Honestly, I don't know either. I am not a lawyer. 

But what helps me to keep my feet dry as a redistribution is if there is someone who takes the responsibility for a work, identifies himself as the copyright holder and grants a free license on the work. This is the case for timgm6mb.sf2, and it sounds good enough for my ears, and it only takes 6mb, so it's still the obvious choice for me. 

On 10/28/2023 at 3:21 PM, ceski said:

GMGSx.SF2 side by side with gm.dls and sure enough the samples are identical

Interesting.

I also like SC-55.SoundFont.v1.2b.sf2, the one included with Woof and Nugget (TimGM6mb.sf2) sounds so weird tbh.

1 hour ago, CacoKnight said:

Which one is better in your opinion, this one or RLNDGM.sf2?

Other than GMGSx being louder, they sound pretty much identical (at least to my ears)

Here, listen to these recordings made using both soundfonts (and a third one, 4MBGM by Creative, for reference)

https://www.mediafire.com/file/8pyucffo26ghta2/DW-SF2-MID2OGG.7z/file

For this comparison I've chosen two OG doom midis, two modern compositions by the DW community and two demo tracks by Creative/Microsoft.

(this discussion should maybe be moved to a more appropriate topic )

Just wanted to bring up this line in the OpenAL cmake file: https://github.com/MrAlaux/Nugget-Doom/blob/master/cmake/FindOpenAL.cmake#L34C1-L34C71

if(APPLE)
  set(ENV{PKG_CONFIG_PATH} "/usr/local/opt/openal-soft/lib/pkgconfig")
endif()

This line has been really hampering me from building Woof and Nugget from source and I have to go into cmake and manually change this directory to where I actually have OpenAL pkconfig (/opt/homebrew/Cellar/openal-soft/1.23.1/lib/pkgconfig)

I don’t have my MacBook in front of me, but I’m wondering if reinstalling openal-soft via Homebrew might address the issue. It’s supposed to symlink the openal-soft files into /usr/local/opt/openal-soft but it seems like that didn’t happen for some reason. 

Actually I think I know what’s going on. Do you have a non-Intel device? It appears Homebrew uses a different path for Apple Silicon.  

1 hour ago, mikeday said:

Actually I think I know what’s going on. Do you have a non-Intel device? It appears Homebrew uses a different path for Apple Silicon.  

Yes, I have an M1 mac. I even checked usr/local/ and it doesn't even have an opt folder, that seems to be at the root

Does your Homebrew symlink into /opt/homebrew/opt/openal-soft instead? If that’s not it, could you run brew reinstall openal-soft and look for a message telling what PKG_CONFIG_PATH should be set to? Thanks!

Arm64 brew uses /opt/homebrew unlike x64 brew

So both paths need be searched

/opt/homebrew should be prioritized in case both version of brew are installed

I could make a PR this weekend

28 minutes ago, PBeGood4 said:

I could make a PR this weekend

Yes, please!

hi! I need help with running timedemos from windows' command prompt

specifically with appending the results to a text file (useful when running a demo multiple times to get an average score)  

I checked the CMDLINE.txt document without finding anything related

Is there a dedicated parameter similar to -statdump or -levelstat?

I could use stdout redirection but unfortunately it doesn't seems to work:

when running ".\woof.com -iwad DOOM2.WAD -timedemo DEMONAME.LMP" the results are printed out to the terminal 

when running ".\woof.com -iwad DOOM2.WAD -timedemo DEMONAME.LMP > timedemo.out" the putput file will be populated, but the results will also be displayed using a popup dialog (just like what happens when running woof.exe) which require me to click on it to go on with the benchmarks (this makes impossible leaving the machine running timedemos unattended)

/EDIT

actually, I tried out adding the "-NOGUI" parameter and that did the trick!

1 hour ago, liPillON said:

I checked the CMDLINE.txt document without finding anything related

Is there a dedicated parameter similar to -statdump or -levelstat?

We support both -statdump and -levelstat, and they are both mentioned in CMDLINE.txt, but you seem to be measuring FPS? Then those commands are not useful.

Minor bug report:

Woof! does not read MUSINFO lumps properly in that it cuts off a line if it contains a -, eg. entries named D_M29-2 and D_M29-3 will simply become D_M29. This is not an issue with DSDA-Doom, GZDoom, Doom Retro or Delphi Doom.

5 hours ago, deathz0r said:

Woof! does not read MUSINFO lumps properly in that it cuts off a line if it contains a -

Thanks for the report, it will be fixed in the next version.

I think there's an issue with the "max health" field in DeHackEd. I was playing HacX 1.0 (and because it's a lesser known version, I've included a zip with all the files I thought were relevant in case that helps) and medikits were healing me up to 400%, which they don't do in Crispy Doom or the original executable.

4 hours ago, SiFi270 said:

I think there's an issue with the "max health" field in DeHackEd. I was playing HacX 1.0 (and because it's a lesser known version, I've included a zip with all the files I thought were relevant in case that helps) and medikits were healing me up to 400%, which they don't do in Crispy Doom or the original executable.

Make sure you're using the right complevel. Max health in dehacked was interpreted differently in Boom compared to vanilla.

I've already specified complevel 2 in the .bat file. I double checked the menu to see if what you mentioned was a compatibility option I somehow had set incorrectly, but there doesn't seem to be anything like that.

@SiFi270 I can confirm this bug, and have opened up a report on github:

https://github.com/fabiangreffrath/woof/issues/1305