Liquid AI releases LiquidCodec streaming speech codec with weights on Hugging FaceMachine translation

LiquidCodec turns speech into tokens a language model can predict, then converts them back into audio. The roughly 74 million parameter codec has weights on Hugging Face. In Liquid AI's tests, it achieved the best ViSQOL and lowest mel distance across all eight test sets, while NanoCodec scored higher on wideband PESQ in seven of eight.