Abstrait

Effect of Voice Part on the Quality of Children Speech in Dogri Language

Varun Sharma, Padmini Rajput, Randhir Singh, Parveen Lehana

Speech is the most innate and fastest means of communication between humans. Computers with the ability to understand speech and speak with a human like voice are expected to contribute to the development of more natural man-machine interface. For the analysis of speech signal we have carried out the recording of six children speakers (3 male and 3 female) in Dogri language between the age group of 3-6 years. Harmonic plus noise model HNM has been employed as the analysis-synthesis platform as it outperforms almost all models of speech production in terms of important characteristics like naturalness, intelligibility, and pleasantness. PESQ method is used for evaluation of the quality of the speech synthesized from HNM. Mean and standard deviation (SD) is estimated for original and synthesized speech. Effect of different proportion of voice part on the quality and intelligibility of speech signal of children has been investigated at different levels of noise keeping noise part constant. Results suggest that the quality is quite poor at lower levels of voice part but increases gradually until the value of voice part is 50%. However as the voice percentage is increased the quality remains constant afterwards (till v100%). Results suggest that the percentage of voice part plays an important part for the quality of speech. With no voice part the quality is quite poor. Further the results prove that HNM is an excellent model for children speech. Also the worst and best speech quality is not same for male and female children speakers