Hi all!

I'm starting to work on this wonderful world of speech recognition, and more specifically, Spanish speech recognition.

I have started building my own project using CMUSphinx libraries and I have already developed a little spanish speech recongition project using the CMUSphinx acoustic model and dictionary.

I was now wondering whether I could reach better performance using Julius instead of CMUSphinx. And this is the reason why I'm writing this down.

My application just needs some words (10 or less) to work, but needs to be speaker independent.

The first difference I noticed is that CMUSphinx provides their own already build acoustic model, but with julius I need to build my own and it seems to need a lot of work. Do you think is worth the effort of building the model? I have read that the algorithm which internally uses both CMUSphinx and Julius are the same so I would need your advice.

Thank you very much! :-)