1) For workbench, a way to show the full path in window's tittle bar. 2) A very ambitous project would be to add support for voice recogition (I have no idea if there is any open source readliy available for this).
The draw back I've seen in Windows (or probably any GUI system) is that it was designed with the primary human interaction with a mouse / keyboard. If the primary design was voice activation, I wonder what a computer would be like....anyways just a thought about how such a thing could be implemented in AmigaOS. How can it be done better on the Amiga?
I am working professionally on both, Speech Recognition and Speech Synthesis.
Speech Recognition can be from single word recognition to a full blown dictation system. First would be doable, second is too hard to do it in free time, not neccessarily the engine, but the model files for the various languages.
Synthesis can be done; it's quite some work but doable. I wrote alleady a Speech Synthesizer for AmigaOS, I just have not a wide choice of voices yet, I am still experimenting with my own one (German). I got samples from a British English speaker who is really good. But I wanted to create an American English voice first. I got some volunteers, but they are not professional speakers, so the resulting voice would also not sound professional. If there are people reading with an educated voice, please contact me.
Wanderer wrote: Speech Recognition can be from single word recognition to a full blown dictation system. First would be doable, second is too hard to do it in free time, not neccessarily the engine, but the model files for the various languages.
Nice! Main thing for speech recognition would be for voice commands and to enter text (instead of via keyboard)...so "voice commands" sounds more doable than a full blown voice-to-text dictation system
There is Sphinx from Carnegie Mellon University. Sphinx-3 exists in a C/C++ version, and there is a Sphinx-Lite which might be more portable(?). I've looked at it, but as usual, I have had no time to really look at it.
Wanderer wrote: Sphinx is a full blown Recognizer, but who should create the model files? This is too demanding for hobby work.
Yes. But they have one for english, right? You can't sell it, but couldn't you use it for personal use? And I've seen some kind of open source collection of speech data. I don't know how much data they had, but its an interesting project.
Would be nice to do some sort of automatic collection tool, so users of the recognizer could collect data transparently. I have no idea of how much data you would need to get a simple word recognizer? For a large vocabulary recognizer you do need a sizeable amount of data.
well, the thing about OpenAmiga is that everything written should be under the Mozilla Public License, meaning that Hyperion can actually ship it with AmigaOS without breaking any laws. And of course not pay for it, remember what kind of market they are working for...
Not strictly so, the main importance is to not put Hyperion into any legal tangles and make sure that they have unrestricted access to the source. We don't want to repeat things like the narrator device and arexx situation.
Basically the license issue should be looked at from case to case. Closed source is never an option of course, the more restricted the license the less likely that Hyperion will be interested in a particular project. And of course, it's always nice to be able to share source/ideas/examples between openamiga projects without having to wonder too much about incompatible licenses.
@All Openamiga is looking for people to help formulating the license problematics and what stance OA should have on it. You're welcome to either create a thread here or at http://openamiga.org/?function=viewforum&forumid=10