The current documentation is kind of in the validation code, with the introduction of new voices Oct 30, shuffling others off, such as ‘nova’:
src/resources/beta/realtime/sessions.ts
/**
* The voice the model uses to respond. Voice cannot be changed during the session
* once the model has responded with audio at least once. Current voice options are
* `alloy`, `ash`, `ballad`, `coral`, `echo` `sage`, `shimmer` and `verse`.
*/
voice?: 'alloy' | 'ash' | 'ballad' | 'coral' | 'echo' | 'sage' | 'shimmer' | 'verse';
}