A neural audio codec model from Kyutai designed for high-quality speech and audio compression and generation.