Soundwave is a speech-to-text model designed to bridge the gap between speech and text, utilizing a data-efficient strategy and unique architecture. It is trained on 10k hours of data and excels in speech translation and interactive tasks.
