Add to Favourites
To login click here
Slang is a structural approach to sound/signal machine learning, where signals are structured into inter-related annotated parts. This approach enables the signal’s stream to be transformed into a stream of symbols with associated qualifications, quantifications and/or relations that can be used to analyze, interpret, and communicate the signal’s informational content. This article discusses how Slang can be used to develop a language from scratch, by detecting and annotating patterns and relating them to each other from lower to higher levels of abstraction. It also discusses how this approach can be applied to speech recognition.