Through the research we undertook for this master thesis, we developed a prototype, which we believe is the first step, towards building a voice-based controller for a midi synthesizer to be used in real-time. A key component of this research was the evaluation of the different onsets detection algorithms, and their applicability to the prototype developed herein. Another focus of the research was the smoothing of the fundamental frequency estimates, produced by the YIN algorithm. As a result

more »

... the research undertaken, we have found that an onset detection method based on the estimated note changes was more robust to background noise. Our prototype is developed in Max/Msp environment, which renders our prototype open to changes by its user, and make it useful with most software midi synthesizer.