Third party voice recognition on ios

Question

I have the speech recognition working on iOS but it’s assumption that inputs are unrestricted lead it to perform poorly. I’d like to provide a custom grammar to limit the context and improve accuracy.

I see I can do that with pay per use APIs that I would prefer to avoid and the open ears framework that a few people seem to have had dead ends with.

Any recommendations on the best approach?

Thanks