Speech modification for prosody conversion in expressive Marathi text-to-speech synthesis

摘要

During Expressive speech, the voice conveys intended message as well as basic emotions of the speaker. This work focuses on the development of expressive Text-to-Speech synthesis techniques for a Marathi (spoken in Maharashtra, India) language. The Pitch contour is one of the important properties of speech that is affected by emotional speech. This work presents analysis of Emotional and Neutral Marathi speech from pitch contour. Pitch features derived from emotional speech samples with punctuation marks. Presently, question mark and exclamation mark in text are studied to find prosodic information. First, pitch features derived from emotional speech samples are compared with the features derived from neutral speech. Speaker usually tends to put more emphasis on one particular syllable within a prosodic word. We identify such syllable as the core syllable that can be derived with the punctuation marks. The experimental results showed that the proposed prosody conversion, based on pitch modification and prosodic word detection with punctuation mark, can improve speech quality.