You shouldn’t have much trouble using DNS for direct voice-to-text transcription, when you’re dictating directly into you PC using a microphone. You basically train it to recognize your own peculiar intonations and pronunciations. Also it seems to prefer it when one speaks quickly. So neither your speech speed nor your accent (I didn’t know you were from California) should cause difficulties. Nor do you have to listen to yourself in the normal course of dictation.

You can also train it by reading lists of unusual words you use in your writing. So far it’s doing fairly well even with my eccentric character names.

What I haven’t figured out is how to train it for when I dictate on my digital voice recorder. So far it’s proving relatively accurate. Still.

But that only becomes an issue if you’re trying to transcribe recorded dictation.

Also, that’s why I was listening to the MP3s of myself. On direct dictation you watch the words appear onscreen before your eyes. Any mistakes, you can tell it to correct instantly - which also improves the software’s recognition of your pronunciations.

That isn’t possible when you’re transcribing recorded files. Since there still are times when the transcription wanders pretty far afield, the way I figured to proof it was reading the text and listening to the sound files simultaneously.