LITHUANIAN SPEECH RECORDS DATABASE FOR VOICE CODECS QUALITY ASSESSMENT

The English language has had a dominating influence on the advance of telecommunications. With many of the major developments coming from primarily English speaking areas there is the risk that these advances may not be linguistically robust. It is noted that the quality of a speech produced by voice codecs mainly is assessed using samples of English language. However, some other investigations show that influence of language on codecs performance could be noticed. This paper presents the Lithuanian speech records database that conforms to the requirements of ITU-T Rec. P.50 and is dedicated for assessment of voice codecsquality. Experimentally we estimate the quality of AMR and Speex voice codecs using a presented database of Lithuanian speech records as well as American and British English and Arabic speech records databases Supplied in the Appendix I of ITU-T Rec. P.50.