Isolated Words-Recognition Based on Random Forest Classifiers

摘要：

Automatic Speech Recognition (ASR) is basically referred to the pattern classification.However,dynamics of the speech signal has prevented this subject to be put as a simple method of pattern classification.This article introduces a new method for the speech recognition of the isolated words and independent of the speaker and based on the Random Forest Ensemble classification.The proposed method has two major parts.In the first part,the speech signals are converted to a collection of frames,then 12MFCC along with energy and dynamic coefficients will be extracted from each frame,as the features.In the second part,the class of each input speech will be determined by random forest classifier and finally the best class will be determined by majority vote.Validity of the proposed method was tested with the TIMIT database,and the result of evaluation were compared with HMM classifier.

关键词： Speech recognition Ensemble classifiers Random Forest

作者: Mehrdad Attar Mohammad Mosleh Karim Ansari-Asl

作者单位: Islamic Azad University,Dezjoul Branch,Dezfoul,Iran Islamic Azad University,Dezfoul Branch,Dezfoul,Iran Shahid Chamran University Of Ahvaz,Ahvaz,Iran

会议类型: 国际会议

会议名称: 2010 4th International Conference on Intelligent Information Techonlogy Application(第四届智能信息技术应用国际学术研讨会 IITA 2010)

会议地点: 秦皇岛

会议语种:英文

页码: 410-413

在线出版日期: 2010-11-05（万方平台首次上网日期，不代表论文的发表时间）