Telugu Dialects Identification. Using Different Speech Processing Models

      Аннотация: Speech processing is the technique to process and analyze the speech spoken by human beings. The different speech processing techniques are speech coding, speech recognition, speaker verification, and speech identification. ASR is a method to recognize the content of speech uttered by a speaker. Speech identification is a technique to recognize the utterance of the speech that belongs to which language or dialect. Dialect identification is a subdomain in Language identification used to identify the dialects of speech of a particular language spoken by an unknown person. A dialect of a particular language is one form of language spoken in a particular region or environment of human beings where they live. Dialects are different from accents, grammar and pronunciation of the same language. Like other spoken languages, Telugu language (TL) is multiform of different dialects viz., Telangana, Costa Andhra, and Rayalaseema. To identify any language dialects, the standard database is very important. It is a very difficult task to implement a dialect identification system as there is no standard database for dialects and many variations in language.

Ключевые слова: Dialect Identification, Telugu Language, GMM, HMM, PCA, MFCC, ZCR, DNN, Speech Processing