MASSIVE DATA SETS. “The greatest leaps forward in speech recognition and translation software have come not from faster hardware or smarter hand-coded algorithms, but from access to massive data sets of human-transcribed and human-translated words”
https://lnkd.in/dAZH9Vg