AGMSTranscribeSound is a Modula-2 program by Alexander G. M. Smith which plays arbitrarily large IFF and raw sound samples from disk. It is actually a variation of AGMSPlaySound hacked up to be useful for transcribing dialogue from long sound samples. You can use AGMSRecordSound (available separately) to make the dialogue sound samples, up to as many minutes long as you have space on your hard disk drive.
Since AGMSTranscribeSound doesn't load the whole sample into memory, you can play samples longer than your memory size (up to 2 gigabytes). It also leaves lots of memory free for other uses (such as running your text editor in parallel :-). Unlike some other sound playing programs, AGMSTranscribeSound can play from floppy disk (not too useful when doing long transcriptions) without annoying pauses. It achieves this performance by processing the IFF file headers and locating the sound data before starting to play. It also opens all the files before playing so that no time is wasted during playback.
In addition to playing huge sound files, you can specify several samples to be played one after the other by listing the file names in the order you want to hear them (useful for combining phoneme samples to make spoken words or transcribing a series of recording sessions). You can even have the parts on different disk devices. Yes, this program supports floppies too! Note that floppies can only play back data at a rate of at most 10000 Hz (disable any disk caching programs, they waste CPU time when reading files larger than the cache).
The files to be played can be specified with wildcard patterns since this program uses the ARP library. You can even specify the ALL keyword to recursively play all the files in all directories (note that since all files stay open until the program has finished, you may run out of fast memory if you have lots of files, not to mention the memory used for storing the full path name of the files!). I've found and fixed the bug that caused memory trashing with ALL, apparently ARP expects an undocumented parameter to the FindFirst function when it is used recursively. It was documented in 1987 and was then taken out from the documentation. Unfortunately, they kept on using it in arp.library version 39.1. Hence the trashing problem.
How do you get those big sound files? Either use a sound digitizer and an Amiga with lots of memory or use AGMSRecordSound. AGMSRecordSound multitasks while it records arbitrarily long sounds directly to hard disk.
Most of the arguments to the program are obvious. Just type a ? as the argument to get the standard ARP / AmigaDOS command template. A second ? typed at the template prompt will print the program credits (now in AmigaDOS 2.0 compatible ANSI text). The Verbose option also explains in more detail what the current settings are. Ok, maybe they aren't obvious. I'll describe the parameters in detail near the end of this message.
The program supports audio priorities and recovers gracefully from stolen audio channels (waits until a channel is free or the user types control-C). In fact, if you are playing more than one sample, it will backtrack over multiple files to the one which was being played when the sound channel was stolen (one of the advantages to keeping all the files open until the program ends). By the way, I've set it up to give preference for using the right hand sound channels before the left hand ones (sorry, I haven't had time to do Stereo, have a look at several other newer player programs (OmniPlay, DSound, etc) which do).
Bug Fixes and New Features
I've found a bug in the M2Sprint IFFR.mod module for reading IFF files containing CAT objects. Fortunately, M2Sprint includes the source code for the libraries so I was able to fix it in the December 27 1990 version. I like that. The problem comes from not using the listContext that was carefully set up to read the CAT. Instead, they use the clientContext parameter to get the address of the procedure to call. For CAT objects, clientContext is nil so some random memory address gets called. Where did I find it? AmigaVision uses CAT objects for containing the elements of your program. Now if someone out there has IFF files with LIST or PROP objects, I'd appreciate a copy for testing... Never mind, AGMSMakeFilm generates that kind of file. By the way, if you want to do video off a hard disk with a stereo sound track, try out the AGMS*Film series of programs.
Also fixed (a feature really :-) is the file closing algorithm. Files that were examined but have no sounds are now closed after examination rather than at the end of the program. The memory used for their path names is also freed at this time. This should make it easier to play all the sounds on your hard disk without running out of memory.
The NoWait option was added in the September 6, 1992 version after James Atwill requested it.
Faster disk access (got rid of some unnecessary seeks - really huge sound files will be examined much more quickly), 2.0 compatible ANSI text and swapping of control-E and control-F were added in the June 13, 1993 version.
Jamie Hellstrom requested a version for doing transcriptions in June 1993. The result is a variation of AGMSPlaySound called AGMSTranscribeSound. The control keys got rearranged yet again to handle transcription functions. I also found the AmigaDOS 2.0 lockup bug, AbortIO sometimes hangs! You tell it to abort an audio write request and it just sits there. Fortunately, stealing the channel will wake it up (run another program that steals channels, like most mod players). I've worked around it by doing a CMD_FLUSH before doing any AbortIOs.
If you have any feature requests or find any bugs, please send me a message. I'm on several of the Ottawa BBX's (a BBS written by SteveX) and my commercial info service names are listed in the program's second help message (agmsmith@BIX.com, 71330.3173@CompuServe.com).
AGMSTranscribeSound is FreeWare. Copyright (c) 1993 by Alexander G. M. Smith. That means that you can use it freely, can't blame me for anything that goes wrong (there are probably a few bugs left), can't claim that you wrote it, and you shouldn't expect more from it than you paid me for it.
This argument to the program is a list of file names. The files will be played in the order listed. Each name can include wildcards (standard Amiga plus ARP style), if it selects more than one file, that group of files will be played in alphabetical order. If you want to repeat a sound sample, just specify its name several times.
This parameter specifies the number of chip ram buffers to use for the sound. You need at least 2. While the system is playing one buffer, the others can be filled from disk. The default is 2 (you don't really need more for transcription of long files - you aren't switching between lots of tiny files). Try the verbose command switch to watch the buffers being filled and to see how changing the number of buffers affects performance.
This parameter sets the size of each chip ram sound buffer. The default is 10000 bytes, the maximum is 131072 bytes. So, if you have 5 buffers and each is 10000 bytes then you will use a total of 50000 bytes of chip ram. There is also a small amount of fast ram used for each buffer (message records and other small things). For straight playback performance, use large buffers. That is because DOS can read a large chunk of data faster than several small chunks (less overhead). For transcription, a smaller value is better, to reduce the lag between your commands and the time you hear the result. The default value of 10000 is good for transcription.
This parameter controls the playback rate for raw sound samples (in other words, when AGMSTranscribeSound can't figure out the speed which the sample was recorded with). It is in units of samples per second, or bytes per second since each sample is one byte long (compact disks have 2 byte samples and thus sound better). The default is about 10000hz.
It is internally translated into a code value for the hardware. Since the hardware doesn't have all that many code values, only a few frequencies are actually available. A frequency near the one you specified will be picked.
Of course, with the variable playback speed added to do transcription (slow, fast and even backwards) the Hz setting is only accurate when playing at 4/4 speed (use the verbose option to see what the play speed is, printed when you change the speed using the control keys).
If this switch is specified, IFF sound samples will be treated a bit like raw samples. The IFF header information about the playback frequency and volume level will be ignored; the default Hz and Volume parameters will be used instead. This is useful if you want to hear an IFF file at lower or higher pitch than it was recorded with. Also, the full chunk size will be used instead of the size specified in the IFF sound parameters for one shot sounds. Unfortunately, it still doesn't help play back mangled IFF files (you will have to mangle the first few bytes so that it isn't recognized as IFF).
This specifies the volume level of the playback. 0 means quiet, 64 is loud. IFF files contain their own volume settings which override this parameter.
When this command switch is used, lots of interesting messages will be displayed. When examining IFF files, the structure and contents will be shown (even for non-sound files like pictures - try it on a picture to see what I mean). All the parameter settings will be shown. The parsing of your wildcards will be shown. And during playback, the program will describe what it is doing (useful for seeing if your buffer settings are good). It also displays the speed and control key commands used during transcription.
No, this is not the task priority. This is the priority that AGMSTranscribeSound uses when it is fighting with another program for control of an audio channel. Larger positive values give more priority. Some programs (like the Sonix player) grab the channels with priority 127 (the maximum). With that high a priority, no other program can take over the channel (other users with priority 127 have to wait). The default value is 90, good for foreground dialogue.
When a higher priority request comes in, the lower priority sound channel user looses the channel (the channel is "stolen"). AGMSTranscribeSound will detect the theft and put in a fresh request for the sound channel. That request will stay pending until the high priority hog has finished making noise (oink, oink!), then AGMSTranscribeSound will be woken up by the Amiga OS and will resume playing from where it was cut off.
This switch will turn on recursive directory searching. Normally, when you specify a directory or one is encountered during wildcard checking, the directory will be skipped. When ALL is used, the directory contents will be examined for sound files.
This one will disable the playing of raw (non-IFF) files. Useful when you use wildcards to play a bunch of files and you only want to hear the ones with IFF sounds in them.
The last switch! Normally AGMSTranscribeSound will wait for an audio channel to become free before playing. If NoWait is specified, AGMSTranscribeSound will terminate if there are no audio channels free. This means that you won't hear (or see, unless Verbose is on) anything if a higher priority sound is playing in all channels. Similarly, if the channel being used gets stolen, AGMSTranscribeSound will try to find another one, and if it can't it will exit rather than waiting for another one. Added on September 6, 1992 to satisfy James Atwill's wish for a Cron cuckoo clock that didn't collect up cuckoos while Experiment-IV was playing SoundTracker modules.
Control-C & Transcription Modes
Control-C will abort the program as soon as it has finished whatever it is doing.
Control-E can be used during the file examining stage to start playing sound samples (stops the file examining activity).
Control-D cycles between transcription modes. The first mode is speed mode (the default). In speed mode control-E slows down the playback by one quarter of the full speed every time you press it. You can slow down the sound to a complete stop. If you slow it down even more, it will start playing backwards (nifty, eh?). Control-F makes the sound go faster in speed mode, doing the opposite of control-E. Note that the Amiga hardware only lets you play sounds up to 28kHz. If your sound sample was originally recorded at 14kHz, you will only be able to double the speed before you hit the limit (speed 8/4, four control-F's away from the default). So, if you want to play sounds at high multiples of their normal speed, you need to record them at a low sampling rate.
The other transcription mode is bookmark mode. In this mode, control-E sets the bookmark to mark the currently loading sound sample buffer. Control-F alternately stops the sound and restarts it from the bookmark position.
A typical use would be to record a conversation of some sort with AGMSRecordSound. Then use AGMSTranscribeSound and your favorite text editor to write down the words. Play the sound file and hit control-E when you hear the start of the section you want to transcribe. Then use control-F to replay it while you write it down in your word processor. Repeat that procedure until you have finished or your fingers hurt :-).
Hopefully the authors of other sound players will be able to do a better job of a transcription program (perhaps using a graphical user interface or a joystick interface (foot controlled switch?)). I've just whipped up this one day hack to give Jamie something to use while the rest of you coders make a deluxe version (I'm busy with AGMSVirtualFileSystem right now).