This paper describes a new hardware architecture for fast Acoustic Model scoring in embedded speech recognition systems by integrating an 8-way data-path with a NOR Flash array. This architecture localizes computation of critical algorithms and reduces decode latency and CPU load by 50% for large acoustic models, resulting in lower word error rates for natural language speech recognition.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.