Mpg123 & AudioCenter

This release has been built in mmpm legacy mode, dynamic audio modules are not available (available audio modules for OS/2 are MMPM and SDL).

This OS/2 port is made by Andrew Alexeyenko andrew@attik.transts.ru, chaos@falt.deep.ru Compiled with pgcc, so requires its runtime dll which is included into the package. You should put gcc29166.dll to any directory listed in LIBPATH or start mpg123.exe from a directory containing the dll. Bugs: it seems that emx multimedia library (or mpg123) do not use DEFAULT audio device, so if there are many audio devices in you system mpg123 will play only via the first one (look at \mmos2\mmpm2.ini)

mpg123 0.59m for OS/2 (mpg123/2 0.11) Modified and ported by Samuel Audet <guardia@cam.org> (C) 1998 Introduction ============ I got sooo fed up with other ports (and their authors) of other MP3 players for OS/2 that I just decided to kludge mine up. Here are the reasons for those interested: - Nobody could (nor wanted to) get a working dynamic priority engine. It WAS quite annoying having to chose between a slowly responsive system and skipping MP3s. but no more! mpg123/2 is here! - Not a freaking single player could tell me the remaining time of an MP3, come on guys! Even the cheapest CD player can do that. - There were no players that could fast forward, rewind, or jump to an arbitrary value in an MP3 file, but it is now a reality! Now we just need get something like XAudio or X11Amp to easily get all that juicy stuff without using the current kludges in mpg123. Hey BTW, XAudio dudes are looking for a FREE copy of Warp 3 or 4 (at least Gilles Boccon-Gibod <bok@bok.net> is). Check their web page http://www.xaudio.com . If you send them something, let me know, I'll guide them to program in OS/2. - No players could smooothly play an MP3 file... I couldn't even do a "dir" without it jerking. But now, you can even play DOS games with the remaining CPU time! (although some games' sound engine don't like it anyway). - And last but not least, mpg123/2 opens the audio device in shareable mode by default. What's the use of having a brain and human ear that can differentiate different sounds in the a stream if you have your sound device locked to play ONE damn MP3!?!? Now answer me... doesn't make sense to me. And you can also choose which sound card to use. The EXE has been compiled with Pentium optimized GCC 1.0, and linked with ILINK 01.08.r1a_CTC308c (?), but to save space, as usual you will need the ************ *-=LATEST=-* (ie.: the most recent version) ************ EMX runtime libraries. ftp://hobbes.nmsu.edu/pub/os2/dev/emx/v0.9c/emxrt.zip GCC290.DLL is a new DLL like the EMX ones, so you might as well keep it. I'm looking for some information on how to program an equalizer (the one in mpg123 source code is a sound mangler). And NO, I don't want to do anything what so ever for 8 bit sound support. Get your ass up and buy a $10 16 bit sound card down the street. Notes on features ================= Read mpg123.txt for things not discussed here. Dynamic priority boost ---------------------- What it does, is that it checks if the second ahead buffer is filled, if it is not, it boosts the decoder priority. When the decoder realizes it has the third ahead buffer filled, it drops to normal priority. The second buffer ahead is important here. During a boost, priority should normally be in time critical class, where the decoder can get all the CPU attention it needs. Up there, if the CPU is fast enough to decode and fill the second ahead buffer before the first ahead buffer is played (which is always the case if your CPU can play MP3s in realtime in the first place), you can easily imagine it's impossible for mpg123/2 to skip in those conditions. I have played *2* MP3s on my P150 and watched a "dir /s" smooothly and happily scrolling in the foreground without either MP3 skipping... I think that's pretty convincing. I know hardware level bottlenecks which could make mpg123/2 skip, but they are the same ones that can make a Zmodem transfer give "CRC32 errors", and a PPP connection complains about "Invalid FCS": - You have a polling device (like for example old old 2x CD-ROMs, and PRINT01.SYS without /IRQ). - You have a video card that supports "automatic PCI bus retry", but your motherboard stops processing during those retries. Disable that feature. - You have an ATI Barf64 and you are experiencing a bug in the video driver using software mouse pointers (ie.: colored mouse pointers). I recommend the default values that are class Time Critical delta 0 for boost and Regular delta 31 for normal. The former leaves higher priority tasks like multimedia do their job. The latter permits mpg123/2 to have priority over hogging DOS and Win-OS/2 sessions while leaving foreground OS/2 application responsive. The format for the priority parameters -B and -N is [class number][delta number] class number in order of priority: 1,2,4,3 delta number varies from -31 to 31 For example: use normal priority Foreground delta -31 [C:\]mpg123 -N 4-31 If you don't want to use dynamic priority boost, just set -B and -N to the same value. Buffers ------- You can specify the number of audio buffer to fill in advance. The default is 32, and since each buffer is 16KB, this gives about 3 seconds decoder independancy to the audio engine before a priority boost or a skip occurs with 44.1kHz, 16bit and stereo output. The recommended minimum is 5, and the _bare_ minimum (ie.: if you don't want to) is 4. Rewind, Fast Forward, Position Jump ----------------------------------- They are still only kludges. Especially the rewind. I only used some already in place code in mpg123. Well they work, so it's better than nothing for now. You will have problems with rewind around the 0 seconds. Try 5 or more seconds. The Jump to accepts seconds, or minutes and seconds separated by a colon. ie.: Jump to: 1:50 or Jump to: 110 Since these mechanisms act upon the decoder, the more audio buffer you have, the greater the latency. With 5 buffers, you have less than half a second latency. The Played, Remaining and Total timers -------------------------------------- None of them relies on the system timer. It only counts the number of frames that have passed by (that is, played by the audio device, so in a certain sense, the audio device determines where we are). The calculation seems to work, if they don't on some MPEG Audio streams, oh well... It should also be noted that the Total time is updated each frame, so if you are currently playing an MP3 being d/led, the total time will increase with it. These also act on the decoder, so it shows the amount _decoded_ and not _played_ per se. Of course, they make mpg123 use a tiny bit more CPU time... MP3 filenames and playlists --------------------------- Ok, a nice little feature requested by William Young <young@phonet.com>. You don't have to specify ".mp3", mpg123/2 will automatically append it if it doesn't find the file. A trick to easily generate playlists: [C:\MP3]dir /f/-p > playlist Use /s for all subdirs. Well have phun! Expect to see some nice MP3 player around (maybe pm123?). Legal stuff =========== This freeware product is used at your own risk, although it is highly improbable it can cause any damage. If you plan on copying me, please give me the credits, thanks. Oh and SDG you are not allowed to use this sound engine until you implement all the features in here AND LET ME TRY IT BEFORE YOU DISTRIBUTE IT. That should do it. :) Contacting the author ===================== Samuel Audet E-mail: guardia@cam.org Homepage: http://www.cam.org/~guardia IRC nick: Guardian_ (be sure it's me before starting asking questions though) Talk Client: Guardian@guardian.dyn.ml.org Snail Mail: 377, rue D'Argenteuil Laval, Quebec H7N 1P7 CANADA

NOTES ----- I have included the CHANGES file from the unix mpg123 code. Some of these changes are not represented in the OS/2 version, for a variety of reasons. Most notably, http support is not currently functioning, and the new buffering system was removed (as it was redundant). The commandline switches -have- changed, so type "mpg123" by itself to view the new options, or read the included mpg123.txt file This is, at best, an "unofficial" port. Please don't email the original program author(s) with questions. This port was compiled with VAC++, and DART support. In fact, it -requires- DART. If you are using an early version of Warp 3.0, you should be able to find the necessary DART .dll files in the Netscape add-on package. This port is very much a "work in progress". The unix varieties of this program are continuing to grow in versions, and I'll be making some effort to keep up with the changes. If you are running mpg123 with low CPU power (or trying to), I cannot recommend the "-m" option enough. Locally (p133), running normally mpg123 takes up 35.1% CPU. Running with "-m", that drops to 18.2%, a fairly significant drop. email at bhruic@servtech.com Craig - 05/05/97 v3 - Started making some changes in preparation for the theoretical PM version - Pause/Resume commands added - Previous/Next commands added (only useful with playlists, of course) - Stop command added - Keeping up with the latest mpg123 *nix code, see CHANGES, keeping in mind that not everything is reflected in the OS/2 port - Seems I accidentally fixed the spaces in filename problem in the last version. Someone just mentioned it to me now, and I re-fixed it (since I had just finished breaking it again :)) - Oopsie. Seems I "accidentally" hardcoded the play rate at 44.1KHz, which was causing samples recorded at slower rates to play faster... It now properly sets the play rate - Except, of course, finding that made me realize that it wasn't switching sampling rate if there were different rates in the same playlist. After a great deal of struggling, I've figured out a "kludge" to do so, but it doesn't work as well as I'd like. But a switch from 22KHz->44KHz->22KHz worked, so I'll leave it at that for now. - Added wildcard support. - Added recursive subdirectory searching (-e). This in combination with wildcards will allow you to play all the mp3's on your HD (if you want) without making a playlist. Note, if you have a large HD with a lot of files, running from root -will- bog down your computer during the search. - Redid the buffer code to fix some problems and to be compatible with the commandline -b flag. Somehow the buffer changes have caused the program to use 3% less CPU on my system. No idea why, but I'm not complaining. :) - Got the 8 bit code from the latest version and patched it in for the OS/2 version. 8 bit code has been tested, and seems to be working. - Made some other minor adjustments not really worth noting (although as a group, it seems they are, since I just did :))