I've looked for a long time for a program that can play the sample data from stdin but failed; as a result this one was born. Besides, I needed a simple sound output library that uses DART, so this program was used both as a testbed and a usage example for it.
Its main usage for me is to serve as a output device for "sox" program, maybe you can find other usages for it. Within combination with sox it serves as a great sound file player, allowing even for such effects as real-time flanger, vibro, reverb and so on - you can use just any effect that sox can do. The only limit is that your CPU should be capable of doing the effect in realtime; otherwise devaudio will detect the end of input stream and quit.
The basical usage is simple: when devaudio.exe is run without parameters, it assumes default parameters of signed 16-bit PCM sample format recorded at the rate of 44100Hz, two channels (i.e. stereo) and uses default waveaudio device for output. The data read from stdin is "copied" as-is to the output device, and that's it.
You can change any of the default parameters so that output parameters will match those of the input stream. devaudio recognizes the following options that can be used to change the way audio device interprets input data:
-d# --device=# Choose audio device index. By default devaudio uses
device with index 0 which is the "default" waveaudio
device in system (the one that has the "default waveaudio device"
checkmark in the Multimedia Setup notebook)
-b --bytes Set input data format to one byte (8 bit) per sample
-w --words Set input data format to one word (16 bit) per sample
This is the default setting.
-r# --rate=# Set input data sampling rate. By default devaudio
assumes 44100Hz.
-c# --channels=# Define number of channels. By default devaudio assumes
two channels, i.e. stereo data.
-f# --format=# Define input data format.
Supported formats: pcm, adpcm, ibm_cvsd,
alaw mulaw, oki_adpcm, dvi_adpcm, digistd,
digifix, avc_adpcm, ibm_adpcm, ibm_mulaw,
ibm_alaw, ct_adpcm, mpeg1.
By default devaudio assumes pcm data format;
other data formats can be supported IF the hardware
supports it; if hardware does not support a specific
format, devaudio will report a error and halt.
-p --progress Display progress bar as sound is playing. Unfortunately,
this is not too useful as it does not work on pipes (pipes
are not seekable, so you cannot determine "file" size).
-s --switch-sign Switch sign. If you have an unsigned stream of data and
the format you use supposes signed data, or vice versa,
you can use this option to switch sample sign. I.e. if
data was signed it becomes unsigned, but if it was
unsigned it becomes signed. This switches makes devaudio
consume a bit (a *little* bit :) more CPU.
All switches have both "long" and "short" form; they are equivalent, i.e.
"-b" is equivalent to "--bytes"; "-c2" is equivalent to "--channels=2" and
so on.
Usage examples:
The versions of SOX that can be found on all OS/2 sites have one consistent bug: when writing data to stdout they do it in text mode, i.e. each "\n" is converted into a "\n\r". This leads to awful results. I've included a precompiled version of SOX 1215 that doesn't have this annoying habbit. The force SOX use stdin and stdout in binary mode you should add the following lines:
#ifdef __EMX__
// Set stdin and stdout to binary mode
_fsetmode (stdin, "b");
_fsetmode (stdout, "b");
#endif
right after main () in sox.c. These are almost all changes, except that I've
changed a bit the makefile to use OMF object file format instead of a.out and
added the -Zbin-files during linking phase instead of using "rb" and "wb" open
flags; these changes are obvious and are not listed here.