search

jhudsl / ari

:dancers: The Automated R Instructor
https://jhudatascience.org/ari/
Other
146 stars 38 forks source link

Integrate open-source Text-to-Speech synthesizers #32

Closed The-Pezman closed 3 years ago

The-Pezman commented 3 years ago

ARI is a great package, but I don't love the fact that its speech synthesis is dependent on Amazon. Not only does it complicate the setup process, it makes ARI less open source.

Adding an option for open-source speech synthesizers like eSpeak, Mycroft and/or MARYTTS would address this issue. They can all be easily downloaded and compiled on any OS, and have open licenses. Granted, I don't know how easy they are to call from R, but since they're written in C, I don't imagine it's terribly difficult.

Not all the voice options are great, but some are certainly adequate (at least in English). Adding additional mbrola voices could potentially help.

muschellij2 commented 3 years ago

We use the text2speech package. As per https://github.com/jhudsl/text2speech/issues/3, these can be incorporated only if an R package is already created. "I don't imagine it's terribly difficult." - if that's true, then a welcomed PR to text2speech would allow us to use these.