Hi Everybody!

I need to create a project that converts the human voice into the written word, something like this. Can Sphinx4 do this job? The project is for Italian and English language.

Thanks!

Yes, Sphinx4 can do that. You can try a similar technology available open source here:

http://kaljurand.github.io/dictate.js

or here

https://www.cloudasr.com/demo

For Italian you will need quite some work to add support for Italian language though. You need to collect significantly large speech database, about 500 hours of speech.

Many thanks for your response, I'm going to look the projects (i saw that one is using kaldi). What if I need get only some main concept from a speech, the process to train the model is the same but i need a smallest speech database?

Thanks again