It is common that after recording, we would like to correct mistakes made prior to recording. One of these mistakes is the incorrect pitch. We can choose to re-record the material over and over again, until we are satisfied with the result, or we can choose to use signal-processing methods after recording. In my thesis, I focus mainly on poliphonic choir records, and synthesized signals carrying the same set of attributes.
This method is rarely used in practice, mainly because it can only be applied on a certain kind of input signal. Monophonic pitch-correction, however, is a frequently used tool in modern music. It is usually applied on vocal tracks to correct the pitch of the notes, or in some rare cases, it is used with the intention of making the presence of this effect heard through its unique side-effects. Pitch-shift algorithms used for correction can be found in practice as well, especially used for modifying the character of the input vocal, however, in this case, as opposed to pitch-correction, a constant shift rate is used. In pitch-correction, the shift rate is changing, depending on the pitch of the input signal.
The objective of this method is the realization of a correction that does not modify the unique set of attributes of the input signal, keeps its natural sound, while achieving an output with correct pitch. The output signal must be the same length as the input signal.