|Abstract||Transformation of polyphonic audio in terms of pitch shifting and time stretching is often desirable and sometimes necessary in order to correct a mistake in the performance of the musician or in order to experiment with different chords or durations, that can be produced by transforming accordingly the given audio piece. Transparency in such transformations should be a key feature that would oer persuading results. In this Thesis we implement a framework that carries out such processes by embedding different modules and techniques such as filter-banks, pitch shifting algorithms, and harmonic gain compensation blocks. We apply transformations between different chords on several instruments of different characteristics , and we measure the quality of the transformed sound. Furthermore we mention some inherent vulnerabilities of the system and we propose ways to overcome them.