speechd modules

[Tim Cross]

Hi All,

I'm new to speechd and have to say I like the concept and
implementation as far as I've seen it so far. 

I'm thinking that in my spare time, I wouldn't mind adding some new
output modules for speechd. In particular, I thought I might look at
doing some modules for the software dectalk and possibly Cepstral's
TTS engine. However, I thought before I begin to look at this, I'd see
if anyone else was working on anything with these software synths as
I'd prefer to work with someone rather than reinvent what they have
already done. 

I realise speechd is pro open source and neither of these synths are
open source. However, I also feel we need some modules with high
quality speech and there is some validity in supporting commercial
efforts with TTS, especially if the retail price is reasonable. 

Just for some background on me. I've been working as a programmer for
about 16 years and a Linux user for the past 10. I lost my vision
about 7 ytears ago and have been using emacspeak since then. I'm very
interested in the speechd-el as an additional alternative
approach. While I totally support and commend T.V. Raman for the work
he has done, I feel emacspeak has grown a little bit large and
difficult to manage. In fact, I've been tossing up the idea of
re-writing parts of it to simplify parts etc. However, speechd-el
could offer an alternative approach where we could try implementing
some of the nice features of emacspeak in a relatively simple way
without major re-writes etc. I'm not trying to be critical of
emacspeak in any way - the issues I have with it are just a reflection
of the type of issues you will find in any system which has grown and
evolved over a long period - especially one that has, to a large
extent, pushed the boundaries of traditional speech systems. I'm not
interested in making speechd-el a clone of emacpseak, but there are
some features in emacpseak which I think wold be valuable additions to
speechd-el. 

One area where emacspeak is very much lacking is in its interface to
speech synthesizes. There are few support synths and its a bit tricky
to add new ones. This is in particular what I like about the speechd
approach. I think the abstraction being used is a great idea and
offers a lot of possiblities. 

Having said all of this, I should point out my time is quite limited
and it will take me a while to actually make much of a contribution to
either speechd or speeched-el. I made a slight tactical error in my
employment about 8 months ago and took on a 2 year position managing a
reasonably large data centre. This is now taking most of my
time. However, I miss coding and like to still spend some free time
hacking away at code and this is why I'd like to work on some project
- albeit very slowly. 

so, I'd like to here if my ideas and desire to contribute will fit
into the philosophy of speechd and if anyone else is working on either
software dectalk or Cepstral support for speechd. 

Tim