Hey! From the way I see it, to change a human voice you need more than simply changing pitch, frequency, speed, etc. Humans evolved to notice and feel strange even the slightest abnormalities in the human voice. Therefore, I believe that this is a task that will need artificial intelligence to be complete
My idea is: Start by building software that will take real-time voice and process it according to the parameters the user defines. This means, changing all the attributes you mentioned (chiefly frequency, harmonic structure, intensity...) and some more. However, this is not enough, this is only the first step. This change will be purely algorithmic, in the sense that you change the sound wave exactly as you define it. However, after this, you'll have a second layer, that you can turn on or off, and that I believe will be the game-changer: The AI model.
The voice produced by the first layer alone will probably not be completely realistic, as some intonations, and ways of speech may not sound well in certain parameters. Therefore, I will train an AI model that will try to "Humanize" any voice given to it. This will make sure that the voice we get not only respects the parameters we defined but as important, actually sounds human-like, instead of robot-like
If you want to discuss this in more detail, and want an explanation now how such an artificial intelligence model will be built, send me a message and let's discuss this further.