Cutting ASR Inference Cost with NVIDIA MPS on Amazon EC2

When an ASR pipeline is pushed to production, the interesting question is not only how fast it runs,...