is the pronunciation for aorist passive
 

with theta epsilon or theta eta different? Or is it simply an orthographical change?

Future middle voice? eh I'm dying.

I doubt anyone really knows. There are various theories as to how ancient Greek sounded (the accents are a remnant of the pitch, not stress, associated with each syllable), but nobody knows for sure. Different schools teach different conventions. For instance, I have met people who insist on pronouncing the omega-with-iota-subscript as a diphthong 'oi,' while most people just pronounce a long 'o.'

Vowel changes in ancient Greek are a combination of euphony and dialect. Since ancient Greek was primarily a spoken language, the Greeks avoided certain sounds that they found unpleasant. When they started writing them down, they approximated as best they could.

As for the H (eta), it's a relatively late introduction into Attic Greek. Before the 420s BC, they just used A (alpha) for both vowel sounds. Doric maintained its alpha spellings.

Basically, there were three main dialects of Greek: Ionian, Aeolic, and Doric. Each of these developed various sub-dialects, such as Attic from Ionian Greek. They're tricky to read at first, but once you figure out how they sounded, they're not so bad. Koine Greek, I suppose, can be considered a derivative dialect as well, but it's a simplified GFL (Greek-as-a-foreign-language) version.

The vast majority of ancient literature is in Attic Greek, which is why that's what the textbooks teach. If you're interested in the different dialects, check out Greek Dialects by Carl Buck.

Sorry for the long, pain-in-the-rear response. Learning forms is really tough, but in the end you just have to be able to recognize them (unless you want to do some composition). The theta, in general, indicates a passive voice in a verb. That's usually enough to tip me off to its form.

Smyth doesn't even comment on why the difference, other than to note that epsilon appears in aorist passive moods other than indicative, imperative, or infinitive (590). Even in these latter cases, however, I think that epsilon ends up combined with an iota (usually). In the case of verbs like tithemi, whose plural forms end with a penultimate epsilon, then yes, I think the pronunciation would be different, since the accents are on the 3rd to final syllable (proparoxytones? hell if I remember). As for HOW they should be pronounced, well, that's a tough one. I don't know for sure. I just always used the typical classicist pronunciations, but who knows how accurate they are. These could be a lot closer in pronunciation than classicists are taught.