Hibiki is a Rust-based model for high-fidelity simultaneous speech-to-speech translation, adapting its flow to produce real-time translations in chunks, with optional voice transfer and text translation.
