Speaker Independent Continuous-amharic Speech Recognizer An Experiment Using Hybrid (hmm-ann) System

Automatic speech recognition is a part of natural language processing whichrndeals with making computers hear human speech and to take action according to whatrnthey heard. In line with this concept, Speaker Independent Continuous Amharic SpeechrnRecognition is investigated to check the results that can be reached by an ANNIHMMrnhybrid approach.rnTo implement the Amharic ASR system , 100 speakers, each speaking ten differentrnsentences have been modeled with the help of the CSLU Toolkit . The model is constructedrnat a context dependent phoneme sub-word I eve!. A promising result of 78.56% wordrnaccuracy and 44.07% sentence recognition rate has been achieved for speaker dependantrntest and 74. 28% word accuracy and 43.70% sentence recognition rate for speakerrnindependent test. These are the best figures reported so far for speech recognition forrnAmharic.

Get Full Work

Report copyright infringement or plagiarism