The development of acoustic models for command and control arabic speech recognition system

Abstract

The paper presents an acoustic training system for building acoustic models for a medium vocabulary speaker independent continuous speech recognition system. A speech database is constructed to train the acoustic models. The acoustic models are constructed, trained. A test set database is constructed to test the accuracy of the acoustic models, also 4 language models of two main types: bigram and context free grammar were built and used in tests. Our results show a 5.26 YO and 2.72 % word error rate for 1340 and 306 words bigram based language model respectively. Our results show also 0.19 % and 0.99% for 1340 and 306 words context free grammar based language models respectively.