Neural audio codecs: how to get audio into LLMs

Kyutai unveils today the first voice-enabled AI openly accessible to all [pdf]