A cognitive method for melody stream transcription

This paper describes a method for analyzing tonality and transcribing melody stream automatically. Tonality plays the core part in organizing music structure. By making an investigation into cognition and musicology, we propose an approach to transcribe major component of polyphonic music from the point of view of tonality. Our method focuses on music cognition instead of signal processing. Basing on our former work on key/chord recognition, we introduce a cognitive feature called auditory saliency, which contains both statistic modeled information about human´s acoustic attention and psychology measured data on human´s musical perception, to recognize the melody stream. Constant Q Transform(CQT) and a temporal ANN are also used in our framework. Our proposed method has been tested objectively (F-measure of pitch detection is 0.63) and subjectively (around 90% of segments are accepted), and shows some inspiring results.