speech processing

Voice Conversion System

The system aims to generate sentences by utilizing voice properties of target speaker. Project encompasses inter and intra gender voice conversion functions for source and target speakers i.e. male to male, male to female, female to female and female to male. The project is developed by using State-of-the-art techniques to deal with voice properties of unknown speaker as well. Currently, this system is employed to English sentence and for Urdu dataset the work is under progress. The interface for the system is shown below.

Figure 1. Voice convertor system takes two audio files as input; one from the source speaker subject and the second from the target speaker subject.
Figure 2. The output file is generated for the target input by the system