Using full duplex with OSS
==========================

There are two _alternative_ ways to use full duplex with OSS:

1) The original full duplex API permits one process to use the device in
both directions simultaneously. However this method requires an application
that is written to use this method. This alternative is recommended when
writing a new application that requires full duplex.


2) With some soundcards there are two separate audio devices (visible
using cat /dev/sndstat). In most cases the second device is a "shadow"
device but in some cases there are two functionally independent devices
available (usually the second one has only output capability).

With two devices it's possible to use applications like speak freely.
Use the first device (usually /dev/dsp0 and /dev/audio0) for recording
and the second one (/dev/dsp1 and /dev/audio1) for playback. For example

cat /dev/dsp0 > /dev/dsp1

Note! Under some operating systems (Linux) it's possible to open /dev/dsp
twice (once for recording and once for playback). This makes commands
like "cat /dev/dsp > /dev/dsp" possible. However there may be some
unpredictable performance problems when this method is used.