Comparing amateur and professional musician skill levels using spectrogram representation and simpler statistical parameters

Amateurs playing an instrument have difficulty reaching a professional level without practice and training in the presence of a mentor. We show that the skill level of a person playing a French horn could be distinguished by differences in the tone shape displayed via a spectrogram produced using the short-term Fourier transform. The appropriateness of this approach for comparing amateur and professional timbre was unclear for our target audience: Junior High School musicians. Simpler statistical parameters used for characterizing normal and abnormal electro-encephalographic and vibroarthrographic signals were investigated. These parameters summarize the tone characteristics within a single number, opening up their use within an artificial intelligence context. It was discovered that using the kurtosis measure led to better differentiation of skill levels than the complexity measure, with skewness providing no differentiation.