alice-1.00b
January, 2002.

Things that need to be done...

* Expand and revise currently included language model.
The current language model (in Config/lm/) is just a placeholder
that gets basic commands in place.  

* Fix mp3_module next/previous song scroll bug.  We don't currently
re-window the playlist correctly.  The whole concept of windowing,
i.e. having a start song, end song, and selected song needs to go
away, since it's a carryover from an earlier piece of code that
mp3_module was based on.  

* Clean up the bottom half of the speechio_module event loop.
Functional, but very messy.

* Make Modules/speechrule.py conform.  It doesn't currently read options
from the config file, or follow correct guidelines for log output.

* Add ability to retrieve song by 'track' number in playlist, via a 
command like 'Go to song one four six' , or 'Play song one four six'...

* Scan ID3 tags of all MP3s in playlist on app startup, and do a rebuild
of the LM on the fly, so it includes song / artist names in the vocabulary.
Necessary for the next item.

* Allow retrieval of song by artist and song name:
i.e. 'Play me the song Bookends by Simon and Garfunkel'

* Expand the flexibility of the current speech text recognition and binding
process.  Ideally, speech text entries in alice.bind should be regular 
expression strings that we can iterate through and test for a match when 
a speech text input occurs.  This would allow dynamic info to be included
in speech (i.e. song / artist names).  Currently only static speech is 
supported.

* Make the app easier to shut down.  Currently ./startup spawns programs
which have children, which necessitates going through the process list and
manually killing things.  Not really a priority on production boxes, though.

* Remove Perl dependency (which exists via the sphinx2-alice script that gets
speech recognition going.  Can probably be ported to Python.)