I might lack proper terminology. Imagine you want to create voice recordings for a large collection of sentences for foreign language acquisition. If they reuse of a limited number of words, you could record them and put together to make sentences instead. For the moment please ignore the fact how unnatural that would sound.
Now, some of the sentences would be questions. Those usually end with a rising pitch in English. Does modifying pitch for a part of an audio recording involve complex techniques? Would that require lots of computing power for a modern computer to do that in real-time?