文章基本信息

标题：Emotion Recognition from Hindi Speech Signal
本地全文：下载
作者：Meenakshi Singh ; Parul Khullar
期刊名称：International Journal of Innovative Research in Computer and Communication Engineering
印刷版ISSN：2320-9798
电子版ISSN：2320-9801
出版年度：2018
卷号：6
期号：5
页码：5624-5631
DOI：10.15680/IJIRCCE.2018.0605026
出版社：S&S Publications
摘要：This project is an effort to recognize emotion from Hindi speech signals. Emotions are our life's power and vitality, which are reflected through speech signals and these signals are those carrier wave upon which the projected thoughts are sent to create or modify the reality. Emotion explains much about a person like, its behavior, intentions at an instance, therefore, it can be used in many applications like: in call centers to detect customer emotions, in field of medical (psychiatry), in criminology, in entertainment electronics, in text to speech synthesizer etc. The database used was collected from various speakers belonging to different genders and age group. This work basically focused on eight emotions which comprises of fundamental emotions with some advance emotions and are listed as , Happy, Angry, Sad, Depressed, Bored, Anxious, Fear and Nervous . Classification of emotion is done by analyzing speech signals .These signals were preprocessed and analyzed using various techniques like: cepstral , linear prediction coefficient etc.. In feature extraction of speech signals , the system uses various parameter features are : fundamental frequency, pitch contour, formants, duration(pause length ratio) etc. to form a feature vector and then Knn Classifier were used to classify and recognize those emotions.
关键词：Emotion Recognition;Feature Extraction; standard deviation of fundamental frequency; standard deviation of energy and pitch contour; formant frequencies; duration; zero crossing result