Specify different speech2text models based on language.

[Aran30]

In light of adding a speech2text component (see #25 and #39), in might also be interesting to add the possibility to specify multiple speech2text models based on the chosen language. General purpose speech2text models (that are small enough to quickly provide results) seem to excel in the English language but seem to fall a bit short on other languages. Overall, the idea is similar to creating multiple NLP engines for different languages (see #23).