Event Details
Mel Frequency Cepstral Coefficents (MFCC) Feature Extraction
Presenter: Gorkem Cipli
Supervisor: Prof. Peter Driessen
Date: Fri, July 10, 2015
Time: 16:00:00 - 17:00:00
Place: EOW 430
ABSTRACT
ABSTRACT:
The first step in any automatic speech recognition system is to extract features meaning to identify the components of the audio signal that are good for identifying the linguistic content and discarding all the rest which carries information like background noise, emotion etc. Not for only speech signals but also sound events in the ocean are essential for improving our understanding of natural life and the environment, as well as important questions associated with the marine mammal conservation. Sound monitoring offers the possibility to continuously detect vocalizing animals at much greater ranges and for much longer periods of time. Detection, tracking and classifying of marine mammal sounds can be useful for both minimizing the risks imposed to mammals by human activities as well as scientific applications.
Mel Frequency Cepstral Coefficents (MFCCs) are a feature widely used in automatic speech and speaker recognition systems and can be successfully applied for mammal sound identification in the ocean. This seminar will explain the steps of MFCC algorithm and will try to introduce its potential applications.