Abstract: Using a vision-inspired keyword spotting framework, we propose an architecture with input-dependent dynamic depth capable of processing streaming audio. Specifically, we extend a conformer ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results