Modifying Amazon’s Alexa ASR Grammar and Lexicon – A Case Study

Hassan Alam
BCL Technologies, California, United States

Keywords: speech recognition, human-computer interaction, grammar modification

In this proof-of-concept study we build a tool that modifies the grammar and the dictionary of an Automatic Speech Recognition (ASR) engine. We evaluated our tool using Amazon’s Alexa ASR engine. The experiments show that with our grammar and dictionary modification algorithms in the military domain, the accuracy of the modified ASR went up significantly - from 20/100 correct to 80/100 correct.