Musical transcription requires much burden even for specialists in the musical field, therefore, many trials of automatic transcription have been being done actively. As musical signals are substantially time varying, we have investigated an LMS-based Fourier Analyzer to accommodate time varying characteristics. In this article, the above adaptive algorithm is briefly reviewed. And then, the analyzing system of Kagura musical signals, which convert analyzed results to the MIDI format, is described. Results in MIDI format are converted to musical scores by using commercially available software in order to verify validity of the system. Finally, necessities of GUI (Graphical User Interface) to support Kagura play and its outline are described.