I'd like to artificially generate speech file using some real speech files. E.g., I have 100h of speech files where Bob is talking, and I'll artificially generate speech files that sounds just like Bob, but as if he was dictating a different text. In other words, I have trained a text-to-speech system where the voice imitates someone based on some of their recordings.
What's the license for the generated speech files and do I owe anything to Bob, the original speaker?
I live in the US.
