1 paper · 1 filter
Sameep Chattopadhyay, Alexander Erdmann, Mari Ostendorf
Modern voice-agent systems rely on streaming speech recognition models that operate under stringent latency constraints. This study shows that, due to the limited memory constraint…