A GUI system for speech synthesis through graphical manipulation of spectrograms

A new software system is designed and implemented to synthesize speech and sounds through graphically creating new spectrograms or manipulating existing ones. This system will enable speech researchers to perform speech perception experiments using synthetic speech, modified natural speech or a combination of both. It enhances the ability and flexibility to modify natural speech for testing purposes in a user-friendly environment. Speech is synthesized using two different techniques, namely cepstral analysis and sinusoidal generators, to accommodate different purposes for the manipulations. The generated speech was subjectively tested and found to be of excellent intelligibility