Support lingua-go for multilanguage audiobooks

[C-Loftus]

It would be cool to implement multi language audiobooks for textual content used for things like language learning. This is relatively novel and besides for some screen readers like VoiceOver on MacOS I am not aware of any other programs that support this.

To implement this I would use lingua-go and then use some sort of chunk algorithm or look ahead using the teereader to determine the language of the text before it is passed into the piper client. This is probably non too hard to figure out but it is non trivial to do it in a way that is memory efficient.

For this to be implemented there would need to be some way to map a language to a default associated model.

[C-Loftus]

I think I am going to close this as not planned.

Multi language audiobooks seem quite hard and would essentially require a rewrite of the majority of the codebase. It is extremely hard to do multilanguage inference on languages which share the same script. i.e. something like Spanish and Italian is quite hard to distinguish. Something like Chinese and English would be feasible, but I am unclear if it is a good idea to support something like that with only certain combinations of languages.

Additionally, I think it might be a better solution to implement this in the text to speech model directly so you don't need to chop up each segment and then stich it back together. For something like a language learning textbook, you'd need to do a significant amount of small audio generation which is much slower than having long runs of the same language all at once. (unless you do something sophisticated to generate everything of the same language and then split it back up and put it in the proper order again