Published on February 6, 2025, Hibiki is described as a decoder-only model for simultaneous speech translation between French and English, trained on synthetic data.
Published on February 6, 2025, Hibiki is a decoder-only model for simultaneous speech translation between French and English. According to the description, its synthetic training corpus uses MADLAD translations, the authors’ text-to-speech (TTS) system, and a simple latency rule.
The repository makes the model available for engineers to evaluate and run. Consult the original repository to check implementation details and results; the supplied description does not specify performance metrics. If you use AI to study or apply the material, avoid submitting personal data or internal documents without authorization.