There are still many cases where MT is not used at all, due to client requirements, for example, so it's not always a matter of choosing between MT and speech recognition. I also think the last question needs a "No" option, and perhaps another one saying "Yes, I now use speech recognition for editing MT output rather than dictating translations from scratch".

Thanks Nora Díaz I agree... have edited accordingly.  Would that change your response?  I'm going to leave the NDA topic out of this poll, although it would be useful to see how much this comes up in the comments.

Thank you, and yes, I have changed it. : )

I use voice recognition for two purposes: 1. it is much faster and 2. it improves my style. I believe that nobody would talk in clumsy phrases like those sometimes produced by written translations.

And yes, MT improved greatly (it is the first time we definitely could talk about a translation, but it still produces nonsense here and there, so very sound proofing is a must. An MT translated text needs more time for review (at higher rates if charged per word :).

It would be great to find in future releases of Studio Voice Recognition specific QA algorithms (searching for similar sounding words or word combinations).