(Courriels de diversion: <cachetteriez@blagueuse-decollement.com> <exceptant@frileux-etheres.com> <suppliciee@tenables-coïncidents.com> <ombragee@pardonnons-malthusiennes.com> <arche@designeriez-cabale.com> <sifflerais@decotes-precede.com> <heteroclites@selliers-decervelage.com> <decapotable@fructifiaient-compatissions.com> <deminerait@prejugez-valoriserions.com> <hypothecaires@dominais-ignorant.com> )
Bonjour Richard et bonjour à tous, Tout d'abord, merci Richard d'être venu sur notre mailing list nous parler du projet GNU Speech. Ce projet est vraiment très intéressant et j'espère qu'on arrivera à le faire avancer. Il s'avère cependant qu'il y a vraiment beaucoup de travail à faire, car si j'ai bien compris il s'agit dans un premier temps de porter le soft sous GNU/Linux après quoi seulement on pourrait s'attaquer à l'implémentation de nouveaux langages ! Pour vous éclairer sur le contenu du projet et donc vous donner une idée de ce qu'il faudrait faire je recopie ci-dessous le contenu d'un message de David R. Hill, PEng que Richard nous a permis de connaitre (merci Richard) : Je pense qu'il serait bon de diffuser le plus largement possible l'info concernant l'existance de ce projet afin de trouver les compétences nécessaires pour le faire avancer. Que pouvons-nous faire ? comment ? vos avis ? ----- Forwarded message from david <david@firethorne.com>-----Date: Wed, 28 Aug 2002 22:23:11 -0700 (PDT) From: david <david@firethorne.com>Subject: Re: RMS recommended me to contact you! To: <info@brlspeak.net> Dear Osvalda La Rosa, Many thanks for your email concerning the GnuSpeech project and your interest in it for use as an aid to the blind. GnuSpeech is intended as Free Software (i.e. under a GPL). You can find out more information by visiting my university web site: http://www.cpsc.ucalgary.ca/~hill and following the menu selection for papers. There's a paper I gave to AVIOS 95 in San Jose which summarises the speech work, and there is also a manual for the MONET system that was used to create some of the databases used for the system that was, at one time, marketed by Trillium Sound Research Inc -- a company we set up to develop the software commercially, but which has been dissolved, following our decision to make it Free Software. The software was called the TextToSpeech Kit and came in three forms: a User Kit, a Developer Kit, and a Researcher Kit. The GnuSpeech software includes all the software for these, plus some in-house tools we created and used for things like dictionary development. The MONET system was central to the development of the English database, and could be used to set up the articulatory database for other arbitrary languages. There is a 70,000 word English dictionary. The system is already partially adapted for French, though, so far, we have not defined a uvular "r" sound, only the five nasal vowels. It would be necessary to define a dictionary of French words, providing pronunciations, plus a set of French letter-to-sound rules for words not found in the French dictionary. It would also be necessary to define additional articulatory rules for sounds special to the French language, though the existing English rule set would provide an excellent starting point. The software was originally developed for the NeXT computer, and makes considerable use of the NeXTStep AppKit and Interface builder facilities, which are currently being ported to Gnu/Linux as GnuStep. The main obstacle to using the software now is that no-one is doing anything to get the software working under Gnu/Linux (though I have some vague promises of help). There are two possible routes to getting the software working under Gnu/Linux: (a) the high road -- port the existing software to GnuStep, the main problem being the immaturity of GnuStep; (b) the low road -- rebuild the syste using C, GTK and things like that to produce a new version based on the old version which can run without the aid of GnuStep facilities, the main problem being that to get the full system running would be a fair amount of work, especially if the intention was to get a French version fully working, because that would require MONET and the interactive tube model interface to be ported, both of which involve significant graphical interface programming (which would be much easier under GnuStep -- I've had reports of NeXTStep applications being ported with virtually no changes needed, thhe difficulties lie in getting GnuStep installed and running apparently. Another research project I worked on was to provide comouter access for the blind, using the TextToSpeech software. The results are written up in the IEEE Transactions on Systems, Man and Cybernetics Vol.18 No. 2 March/April 1988 "Substituion for a Restricted Visual Channel in Multimodal Computer-Human Dialogue" David R. Hill & Christiane Grieb. The work is briefly covered in my paper "Give us the tools: a personal view of multi-modal computer-human dialogue" which has just appeared in the book edited by Taylor MM, Neel F, and Bouwhuis DG "The Structure of Multimodal Dialogue II" (John Benjamins Publishing Co.: Philadelphia/Amsterdam, 2000) pages 25 to 62. The system was called "TouchNTalk" If you find it hard to obtain a copy of either of the papers on TouchNTalk, I'd be glad to mail copies to you. Accessing the web papers should present no problems. I look forward to hearing from you, because I am sure that we can be of mutual assistance. Please don't apologise for your English. I am sure it is better than my French!! All good wishes. david --- David R. Hill, PEng, Professor Emeritus, University of Calgary --- -- Nath ______________________________________________________________________________ Pour mieux recevoir vos emails, utilisez un PC plus performant ! Découvrez la nouvelle gamme DELL en exclusivité sur i (france) http://www.ifrance.com/_reloc/signedell --------------------------------------------------------------------- To unsubscribe, e-mail: biglux-unsubscribe@savage.iut-blagnac.frFor additional commands, e-mail: biglux-help@savage.iut-blagnac.fr