Development of Kannada Speech Corpus for Continuous Speech Recognition

Abstract

The paper presents, development of Kannada speech corpus for speaker independent continuous speech recognition.Speech corpus plays a key role in construction of Automatic Speech Recognition (ASR) and Text-To-Speech (TTS) synthesis.The speech corpora is developed for the age group between 21 years to 45 years.Speech corpus for ASR system is developed by collecting text corpus in which data is recorded corresponding to the text corpus followed by Transliteration (phonetic representation of the text corpus) and finally a pronunciation dictionary is developed.