seliger
commented
6 years ago It certainly is. Of course, life happens and I've not had a chance to circle back to this. One stumbling block that I'd be interested to see worked around is the hand offs for Lex. What I was hoping for was an engine that interprets utterances and returns the values. What they've implemented is something a bit more complicated.
What they assume is that you want to listen for words that describe the intent. Here, I already make the assumption that you came here to listen to the weather. So I'd rather listen for utterances to fill the slots (e.g. a zip code, a city, some sort of geographical marker). That doesn't seem to be the way Lex works.
The only way I can think to work around this is to prompt the user to say something like, "I want to hear the weather." Then Lex can listen for the slots.
Perhaps I've over-noodled on this and the simple way to implement is to have various intents, such as "today's weather," "three day forecast," or "10-day forecast."
If you want to contribute to this, I'm more than happy to take on pull requests. I might have some time over the holidays to mess with this some more as well. No promises though.
Thanks for your interest!
Hey!
Good job! I've realized that Amazon Lex is already available. Are you thinking about launch a new version of this project? I'm interested in helping you to integrate this service to your project.
Regards, Manuel