Voice conversion application (VOCAL)

摘要

Recently, a lot of works has been done in speech technology. Text-to-Speech and Automatic Speech Recognition have been the priorities in research efforts to improve the human-machine interaction. The ways to improve naturalness in human-machine interaction is becoming an inportant matter of concern. Voice conversion can be served as a useful tools to provide new insights related to personification of speech enabled systems. In this research, there are two main parameters are considered vocal tract structure and pitch. For conversion process speech is resolved in two components, excitation component and filtered component using Linear Predictive Coding (LPC). Ptich is determined by autocorrelation. After obtained the acoustic components from source speaker and target speaker, then the acoustic components will be mapped one-to-one to replaced the the acoustic feature from source speaker to target speaker. At least, signal is modified by resynthesis so the resulted speech would perceive as if spoken by target speaker.