This is the result of the Mozilla's hackday after BrazilJS 2012.
The idea is to create a text-to-speech app to run on the Firefox OS.
Due to some restrictions (like audio media and cross-domain requests) we need to setup a small proxy. It is written in ruby, using sinatra and it requires ffmpeg to be installed with vorbis support.
- Install ruby on your machine if you don't have
- Install sinatra ($ gem install sinatra)
- Install ffmpeg with vorbis support (on mac with brew:$ brew install ffmpeg --with-libvorbis --with-theora)
- Run the server ($ ruby server.rb)
It is really dumb right now, and can only read wikipedia articles with brazilian portuguese accent. It is just a proof of concept about a text-to-speech service.
This server uses Google translate text-to-speech. It may take long to run it since it is donwloading the mp3 first. We could implement it using also Mac's native speech-to-text or any other one, which would be faster.