Have you tried speech recognition software in your transcribing?

I will sometimes use it for single speaker files as I mentioned in my article but on a regular basis, I find that doing the transcription myself is generally a lot faster.  I have Dragon Naturally Speaking 10 and I haven’t had the patience to adequately train it.  I do like using it occasionally for dictating emails or writing to my journal where accuracy is not an important factor.  Have you used it or a similar voice recognition software and if so, what are your experiences with it?