Context (archaeology)

Algorithm

Linguistics

Philosophy

Vocabulary

Context model

Biology

Interpretation (philosophy)

Hidden Markov model

Archaeology

Set (abstract data type)

Computer science

Anthropology

Beam search

Paleontology

Natural language processing

Genetics

Contextualization

Sequence (biology)

Object (grammar)

Inference

Sociology

Artificial intelligence

Search algorithm

End-to-end principle

Programming language

Word (group theory)

Spell

Speech recognition

History

In automatic speech recognition (ASR) what a user says depends on the particular context she is in. Typically, this context is represented as a set of word n-grams. In this work, we present a novel, all-neural, end-to-end (E2E) ASR system that utilizes such context. Our approach, which we refer to as Contextual Listen, Attend and Spell (CLAS) jointly-optimizes the ASR components along with embeddings of the context n-grams. During inference, the CLAS system can be presented with context phrases which might contain-of-vocabulary (OOV) terms not seen during training. We compare our proposed system to a more traditional contextualization approach, which performs shallow-fusion between independently trained LAS and contextual n-gram models during beam search. Across a number of tasks, we find that the proposed CLAS system outperforms the baseline method by as much as 68% relative WER, indicating the advantage of joint optimization over individually trained components.

Deep Context: End-to-end Contextual Speech Recognition