I've just created a tag wiki for speech-recognition (
should be under review by now).
I propose that we merge speech-to-text and speech-recognition into speech-recognition because it is the most commonly used term for this kind of technology. (I deal a lot with speech recognition professionally and all the English language forums nowadays call it speech recognition.) FWIW, the Wikipedia entry also lists speech recognition.
I leave the previously proposed merge of voice-commands open for discussion, because while voice commands are part of speech recognition, they can also exist in software not intended for text creation - such as Dragon oder Windows Speech Recognition, but in "pure" command tools like Speakable Items on Mac. However, some people might associate voice-commands with speech recognition for text creation - because you command the computer to write.
My argument in favor or merging voice-commands with speech-recognition is that both categories exist in small numbers, so an overly fine distinction is not warranted. However, the "small-number argument" might not be considered relevant by others.