Any voice can speak any language; however, if you do not use a voice that is native to the language you want the AI to speak, for example, you use a generated voice or a voice cloned speaking English, the AI might have a slight English accent or might drift in and out of similar languages.
For an exact list of all supported languages, please read this article here.
The way the current model works is that you don't select a specific language. Instead, you write in the language you want the AI to speak, and the AI understands automatically. However, since there can be some overlap between different languages where they use similar words and vocabulary but with quite different pronunciations and accents, you should use a cloned voice that was cloned speaking the language with the correct accent as this will ensure the AI has the most context for how to speak something and in what language.
The language is determined by the text, while the accent and pronunciation are determined by the voice itself.
We are looking into developing new technology to facilitate the selection of language, but the way the AI is built means that all of this is still in progress.