Quantifying vocal pitch-classes cross-culturally

Elizabeth Phillips

Although automatic methods for vocal transcription are continuing to improve, the applicability of these methods to diverse, cross-cultural samples is still highly limited. Automatic methods for quantitatively analyzing pitch and pitch-class properties in global song are similarly unsatisfying. Using “objective” methods such as pitch and pitch-class histograms tend to oversimplify the mutable and complex nature of song, whereas using “subjective” methods such as computational melography introduces listener bias and researcher degrees of freedom. In this talk, I will present the methods I have developed for the analysis of pitch-class precision and accuracy across cultures, with the aim of highlighting each step’s relative strengths and weaknesses and prompting discussion about what other possibilities – and hopefully improvements – may be tested in the future.