What Makes Your Voice Unique? Designing and Implementing Vowel Sampling and Processing

What makes your voice unique? What characteristics make up a unique vocal fingerprint, and is it possible to comprehensively and artificially synthesize any human voice knowing those characteristics? Successfully answering those questions is the end goal of this multi-year project. The motivation for this end goal came after seeing the practical uses of speech and vocal synthesis, such as text-to-speech, and vocal impairment aid. These being immensely useful applications, this project sought out to improve current speech and vocal applications, and create entirely new applications by artificially synthesizing any unique human voice.  Currently, a program has been implemented to process vowel sounds, the most basic elements of speech and singing. The program has been designed to sample any person’s voice on a vowel sound, and process it to create many notes varying in both pitch and duration - all in the person’s original voice. This program was created by designing a processing algorithm in MATLAB, a matrix-based programming language. The processing algorithm creates multiple notes varying in pitch and duration using only a single note from a user. This is done using modified linearly spaced vectors, and a floating length central sample processor. While in the past, this project has focused on research and analysis of methods, this year many of those methods are being fully implemented, and new methods are being created. In coming years of this project, methods will be improved and added to finally create a comprehensive and artificially synthesized version of any unique human voice, knowing only the characteristics. The envisioned uses for this project include applications, software, and devices that can use and benefit from any real, comprehensive, and individualized voice.