Yes, you will want 512 of RAM, with room for more, and the fastest processor and front side bus you can get. Speech recognition keeps increasing its demands with each new software generation, and still doesn't have all the power it would like to improve recognition.However, you should also give careful consideration to the sound card and headset you use - that can make more difference than processor speed and RAM to your results. I would tune in to one of the speech lists/newsgroups for advice on what the best performing sound cards are at the moment. Alternatively, go with a good USB headset, or USB adapter for a standard headset, with a sound card built in. In many ways this is the preferable solution, since it means your speech training files can be moved, along with the headset/soundcard, from computer to computer without retraining, and it will work well on a laptop, as well as a desktop computer. If you also have a sound card in the computer, you can use that for uploading/playback, and the USB sound card for recording and listening to your redictation simultaneously. Either send the playback to external speakers, or use two one ear headsets. (Just make sure your left ear does know what your right ear is doing
)