Construction and Evaluation of Tamil Speech Emotion Corpus
- PDF / 208,146 Bytes
- 4 Pages / 595.276 x 790.866 pts Page_size
- 111 Downloads / 202 Views
SHORT COMMUNICATION
Construction and Evaluation of Tamil Speech Emotion Corpus P. Vasuki1
•
B. Sambavi2 • Vijesh Joe3
Received: 15 May 2018 / Revised: 8 January 2020 / Accepted: 18 January 2020 Ó The National Academy of Sciences, India 2020
Abstract Speech emotion-related analysis needs good emotion corpus. The construction and evaluation of two Tamil emotion corpora, one for children and the other for the adults, are described here. Children emotion speech samples are collected from 30 Tamil movies, and the length of the utterances varies from 5 to 40 s. Tamil audio plays are the resources for building the adult emotion corpus. The emotion prosodies are collected, segmented and annotated for the categories of anger, happy, sad and neutral emotions. Observers’ perception test result has been used to evaluate the annotation of emotion. Automatic emotion classification systems have been built using Gaussian Mixture Model and Support Vector Machine. The database is created with an objective of the acoustic investigation of emotion expression in Tamil, analyzing the influence of speaker’s culture and age on emotion expression and investigating the requirement of the need of features unique to Tamil speech on various automatic analysis of speech like emotion recognition, speaker recognition, etc. Keywords Emotion speech corpus Emotion recognition Emotion modeling
& P. Vasuki [email protected] B. Sambavi [email protected] Vijesh Joe [email protected] 1
SSN College of Engineering, Chennai, India
2
Cognizant Technology Solutions, Chennai, India
3
VV College of Engineering, Thisayanvilai, India
Speech is natural as well as the fastest means of communication. Speech conveys information about the language being spoken, emotional state, gender and the identity of the speaker. This work is aimed to build emotional speech corpus, useful for automatic speech emotion analysis in Tamil Language. Emotion recognition has many potential applications like identifying emotion of the kids in various situations like how a game changes a child emotion and analyzing the cause of irritation of the person in psychological analysis and emotion-based information retrieval. The automatic speech recognition can be improved by understanding the emotional state of the speaker [9]. The emotional corpus can be developed in three possible ways: 1. acted speech corpus—developed from the movie or serial clips, 2. spontaneous speech corpus recorded in a real-time environment and 3. elicited speech corpus where professional actors are given the script and asked to act with a particular emotion. Acted speech corpora may consist of music and other utterances of many speakers, spontaneous corpora include noise added in real-time environment, whereas elicited corpora demand a perfect recording environment and professional actors. But for empirical analysis, research works have been are being carried out using all three types of corpora [9]. Emotion recognition is effective only if the recognition engine has been trained with a good corpu
Data Loading...
