Automatic Speech Recognition & Big Data

Automatic Speech Recognition (ASR) is one of the few fields in which Big Data has driven progress before the concept of Big Data existed. The basis of the ASR paradigm lies in the 1970s and little has changed since then other than gradual performance improvement associated with Moore's Law and the associated availability of ever more training data. This video explains this intimate relation and it makes the case that progress does not necessarily comprises breakthrough discoveries (in fact progress, can lock a field in an insufficiently rich paradigm).