[Accessibility] Thinking Further About Speech Recognition

[Christian Hofstader]

Eric wrote a very thoughtful message regarding the ups and downs of 
trying to make an entirely  libre speech  recognition package or 
building a hybrid that uses a proprietary engine that already exists and 
building a libre UI and pre-processor component. The goal would be 
ultimately to replace the proprietary engine with one built by our 
community and carrying a free software license but, in the meantime, 
using some sort of off-the-shelf solution.

This has a very big problem: even if we accepted using a proprietary 
engine  until we could build

    our own replacement, we would be creating a big set back. If there is a
    workable solution - free, proprietary, hybrid, etc. people will not
    volunteer to make a libre replacement unless we can demonstrate how it
    could outperform the hybrid, which will be hard and likely to push
    volunteers into other projects where no analogue exists.

Thus, FSF/GNU cannot endorse a temporary proprietary solution as it will likely 
either postpone or entirely scrap the development of a libre engine that we can 
endorse.

FSF/GNU protects people's computing freedoms and would not want to take 
anything away by using a proprietary solution even temporarily in this case.


So, I propose that we launch a speech reco user interface and  
pre-processor project that is 100% libre. The different sub-projects can 
talk to libre engines and be entirely endorsed by FSF/GNU.

HH,
cdh