Generating speech from text is not very difficult (parsing speech to text is the hard one). You string phonemes together with some sequence rules. e.g. ET's "ph

Generating speech from text is not very difficult (parsing speech to text is the hard one). You string phonemes together with some sequence rules. e.g. ET's "phone home" on Speak & Spell did it with 80's consumer technology. The branding of PC vs. Mac voice could be done independently of whether generation is done on the iPod or in iTunes.