Universal approach for sequential audio pattern search
Róbert Gubka, Michal Kuba, Roman Jarina · 2013
This article deals with universal sequential audio pattern search and sound recognition method. Inspired by classical phoneme-based speech recognition and word spotting systems, where longer speech patterns are formed by sequences of basic speech units, we propose a methodology for creating a finite database of elementary sound models. These models can be arbitrary concatenated into sequences, thus forming a model of the required acoustical pattern or sound event.