We have yet to see any Large Speech Models or Large Multimodal Models that use speech and text as inputs.

It’s only the being but the general process to build this stuff now exists and the value proposition is crystal clear.