Memristor pattern recogniser: isolated speech word recognition

Akshay Kumar Maan, Alex Pappachen James, Sima Dimitrijev

Electronics Letters, 2015

Closed access Metadata (CSL-JSON)
DOI 10.1049/el.2015.1428
Data confidence 85 high
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

A memristor‐ based pattern‐matching circuit for realising a hardware oriented approach to isolated speech word recognition is introduced. As distinct from algorithmic solutions, the approach takes advantage of the memristor to memorise and match speech templates in a fast, parallel, low‐power and scalable architecture. When tested for isolated word recognition by simulations on the benchmark (TIMIT) and Massachusetts Institute of Technology database, recognition accuracies of 94.5% on clean words and 88% on words with a SNR of 5 dB were achieved.