[Flubflow]

For removing hiss, there is/was a good directX audio plugin from Sonic Foundry whereby you sample a section of just the hiss and that is used as a "noiseprint" which is then subracted from the audio. It is not perfect but it is often better than simple EQ filtering. The trick is not to try and remove too much at once with one noiseprint otherwise you sometimes get other strange artefacts in the processed audio (which is probably OK if it is only the speech you are trying to make more legible for dictation).
You can also use harmonic enhancer/exciter type plugins to generate a crisper top end that was not there in the first place. There are also dedicated speech processing plugins too which use a combination of harmonics and filters to enhance the spoken word but these are probably more for fine tuning a good recodring rather than restoration of a very bad recording.