Abstarct: Among the knowledge communication instruments, we can find the ones that are placed in the group of “musical language”. Musical language is a language with its own specific structures and grammars, which have been changing along the past hundreds of years. Getting to know this language needs many years of practice. This thesis aims at recognition and performing of basic musical notes. Along this path, the input image, consisting of a composition that is presented as an image of an input, is taken into account. The preprocessing on the image, specifically correction of the input image angle and noise reduction is performed and the compositions are divided into smaller unbreakable segments. During segmentation, the semantic surges resulting from the picture resolution are modified. After identification of each element and its coordinates, we try to identify the signs by the artificial neural network method. It should be noted that the direction of reading the musical notes is from left to right. Thus, all signs are identified in the order of placement on the staff line. Finally, by the MIDI instrument, we create the MIDI matrix and translate all signs, schedules and spaces into the MIDI language and create their matrix. Finally, we play the composition by and instrument prepared by MATLAB to play the MIDI files.