The Developement of Hindi Sentence Test for Speech Recognition in Noise

Abstract

Background : The aim of the study was to develop a test material in Hindi language for assessing sentence recognition threshold in noise. Material and Methods : The study was conducted in two phases. First phase involved three experiments. First experiment consisted of the collection and recording of the sentence material. In the second experiment, sentence perception was assessed at five signal to noise ratios to check for the equalization of sentence material in terms of intelligibilty. This was done on 25 native speakers of Hindi language with normal hearing sensitivity. In the final experiment, 20 different lists with 10 sentences in each were formulated. The second phase of the experiment involved estimation of signal to noise ratio required to obtain 50% correct sentence identification score (SNR 50) and the test reliability of the lists was assessed. The estimation of SNR 50 was done on 30 native speakers of Hindi language with normal hearing sensitivity. Results : Twenty optimized lists were formulated. Lists were found to be of almost equal difficulty and to have good test reliability in normal-hearing listeners. The average SNR50 was -4.56 dB with a standard deviation of 0.45 dB. Conclusions : The developed test provides a reliable means of measuring sentence recognition threshold in noise for native speakers of Hindi.