NaNo Technology: speech recognition

PhotonQ-Nanobiobots BloodReplication's PetriDish, originally uploaded by PhOtOnQuAnTiQuE.

In previous NaNoWriMo years, I had lots of free time to write at home (or at work, as the case may be). This year, I'm spending up to two hours in the car every day, driving to a job that actually challenges me and keeps me busy. That doesn't leave lots of time to write a 50,000-word novel.

What's the ideal solution? Writing in my car, of course! By dictating during my commute, I could reach my quota before I get home, and keep my evenings open for family time. Therefore, I started researching speech-to-text solutions.

This PC World article lays out the procedure for pairing a digital voice recorder with speech recognition software. The writer sounds satisfied with the results he got using a Sony ICD-SX57 and Nuance Software's Dragon NaturallySpeaking 9, but other reviews have been less favorable. I'm leaning towards an O'Reilly-recommended Olympus DS-2. It's less expensive but still more than I can afford right now. Does anyone out there have a recommendation for doing this on the cheap?