We made state-of-the-art speech synthesis scalable, and achieved a truly remarkable improvement both for the latency and throughput

A Modular Case Study