2003.06.19
Changed the Farnsworth calculations, think I got it right this
time. Thanks to Nake N0NB for help with that. Seems like lots
of people are using cwpcm to create code practice CDs. That's
great! Changed the slow speed (-ss) to 5 wpm with 15 wpm 
characters. Had several requests for that. With the new
Farnsworth code you should be able to change the speed
to whatever you like (>= 5wpm though, I should force the 
character rate to >= 5wpm too).
I use the following to test the speed:
python -c 'print ("paris "*$WPM)[:-1]' | ./cwpcm $SPEED | wc -c
Should get 480000 +/- 4000.

2003.02.28
Added support for MorseMail http://www.seanet.com/~harrypy/MorseMail/
via new executable, cwmm. Same speed switches as cwpcm.

2002.08.01
Added comma, corrected colon ---... (reported by KB8DNR)

2001.10.22
Tested the timing on cwpcm. Running slow, so I did a major
rethinking on all the math. Turns out my original numbers were
all wrong. Wound up reducing the 'half dit' size from 1666 microseconds
to 1177 microseconds. Running 20*"paris " through cwpcm -sx and
out to /dev/audio now takes 60+/-3 seconds, close enough I 
suppose. I tried to correct the split timing as well. 
The problem is that the characters are faster, so the overall
time for a given phrase is less than the requested word rate.
I tried w-(c-w)/w, which was too fast, so I tried w-2(c-w)/w.
Still faster than requested, but I'm going with that for now.
I counted 15 inter-character and word spaces for 36 character
spaces in ".--. .- .-. .. ... " (paris ), perhaps 2.4 is the
proper constant for the above formula.

2001.10.11
Got everything ready for the cwtext 0.90 beta and released
a source tarball. Announced on Freshmeat on Oct 12.
