Date on Master's Thesis/Doctoral Dissertation
Electrical and Computer Engineering
Cleaver, Thomas G.
Musical pitch--Data processing; Sound--Recording and reproducing--Digital techniques
Many mathematical formulas and algorithms exist to identify pitches formed by human voices, and this has continued to be popular in the fields of music and signal pro-cessing. Other systems and research perform real time pitch identification implemented by using PCs with system clocks faster than 400MHz. This thesis explores developing an embedded RPTI system using the average magnitude difference function (AMDF), which will also use MIDI commands to control a synthesizer to track the pitch in near real time. The AMDF algorithm was simulated and its performance analyzed in MATLAB with pre-recorded sound files from a PC. Errors inherent to the AMDF and the hardware constraints led to noticeable pitch errors. The MATLAB code was optimized and its performance verified for the Motorola 68000 assembly language. This stage of development led to realization that the original design would have to change for the processing time required for the AMDF implementation. Hardware was constructed to support an 8MHz Motorola 68000, analog input, and MIDI communications. The various modules were constructed using Vectorbord© prototyping board with soldered tracks, wires and sockets. Modules were tested individually and as a whole unit. A design flaw was noticed with the final design, which caused the unit to fail during program execution while operating in a stand-alone mode. This design is a proof of concept for a product that can be improved upon with newer components, more advanced algorithms and hardware construction, and a more aesthetically pleasing package. Ultimately, hardware limitations imposed by the available equipment in addition to a hidden design flaw contributed to the failure of this stand-alone prototype.
Gault, Travis R., "A voice operated musical instrument." (2007). Electronic Theses and Dissertations. Paper 482.