Seems to me that IBM is way too focused on speech recognition, and ignoring the fact that people can benefit just from text-to-speech. The old viavoice TTS is tiny. Its just one dictionary file, one shared library, and one config file. In total its just a few megs of space. The rest of the 100MB or whatever is all recognition stuff, which is very complex, the vvtutor to train it to know your voice etc etc. I wish they would focus on TTS first and get that out the door. I also wish they'd make the engine distributable. It will never become popular if the engine isn't distributable. As a developer, how/why would I write apps if I can't distribute the engine? Most software companies make their runtimes free and distribute and charge for the sdks. That makes sense to me. I wish IBM would follow this model, but they don't seem to operate that way. I also noticed that they are targeting embedded systems with very expensive OEM type devtools, and also they are winding a lot of their software around websphere, an enterprise platform that costs mucho denero. How does any of this make for accessible apps for the masses? I don't think it does :( -- Doug
