If you do a search in just this forum alone for the term "speech recognition" you'll find a bunch of threads and the upshot is: great for blabbing out lots of text, so-so for recognition at best, and horrible for getting punctuating and formatting. In short, you'll have to do a lot of editing of spelling and so forth.
NOTE: Dragon Home doesn't come with the ability to transcribe audio files. You'll need the full version. Walmart sells the home version, but Best Buy sells full. The price difference may or may not be worth it to you.
ALSO NOTE: Dragon and other trainable speaker-dependent voice recognition software have well-known convergence issues where you think training them is helping, but is in actuality making them less accurate. A dead silent room is absolutely necessary and Dragon isn't lying when they say a high quality microphone properly distanced from your mouth is imperative. Such software works best in a recording studio, not in the average home. A fan nearby one session, ambient sound from traffic another, a slightly off voice due to stress or health the next, without you noticing these little things, will throw it off.
I use speech recognition sparingly these days. It's not really any better than it was in 1984 on 8 bit computers, and to this day, no one has really invented a decent context based system which might aid the selection of the proper words through statistical usage patterns. Although, I believe such is being worked on in smart phone text input systems.