I am developing a ghost hunting game and I am currently thinking about how to add voice lines to the spirit box. In real life "ghosts" are meant to grab words from radio waves to speak giving it a robatic-ish sounding voice. I am thinking about using a text to speech program  to replicate robatic-ish tone.  I have no moral issues using text to speech since there programs that get the consent of VAs to replicate their voices and the technology for it are decades old.  But I do not want to demean  my work by forcing myself to add generative AI tag as my project is mostly my effort and assets that are not mine are legally acquired by human artists. I prefer not adding the tag but I want to be honest to players and I am getting mixed signals on the internet if text to speech is generative AI or not. I am just wondering what the general opinion Itch.io community on the use of text to speech programs. If I am sensing that this community is leaning of yes then I will record my  voice for the spirit box lines (not too much of an issue). 

Not all text2speech AI requires a promp. And not all Text2Speech is AI. (As hechelion said above)

Loquendo and  Balabolka are not AI generated for example.

 In the other hand Bark and CoquiTTS are AI.

In the legal issue, i think there is a CC0 voice bank you can use. In XTTS or (CoquiXTTS) you only need a 10s voice sample of the voice you want the characters to have. It can be even your own voice distorted. Or perhaps since you want 'ghostly voices' you can use your own voice distorted in Audacity with a noise background and a bit of echo effect? IDK what's more easy for yah.

'

But I do not want to demean  my work by forcing myself to add generative AI tag as my project is mostly my effort and assets that are not mine are legally acquired by human artists. I prefer not adding the tag but I want to be honest to players and I am getting mixed signals on the internet if text to speech is generative AI or not.

Good luck with your project.

:-)