Using a Physiological Ear Model for Automatic Melody Transcription and Sound Source Recognition