On the transcription of monophonic melodies in an instance-based pitch classification scenario

In this paper, a computationally efficient approach to transcription of monophonic melodies from a raw acoustic signal is presented. Two different instance-based pitch classification methods are proposed, the choice of which depends on the size of the available training database. In the first method, the conventional K-Nearest Neighbor algorithm is trained on a large database of piano tones and employed for monophonic pitch detection. For cases where the training database contains only one sample from each possible note, a two-step algorithm, combining semi-KNN pitch candidate selection and note sequence tracking, is suggested. It is demonstrated that in the abundance of training data, the KNN algorithm along with a proper choice of the distance measure and K, yields high performance accuracy. Furthermore, the two-step algorithm is capable of compensating for the shortage of data by incorporating prior musicological information in the transcription process.