Subjective listening experiments show an interesting mix of agreement and disagreement among listeners, which can help or hinder analyses depending on their goal. Western staff notation is useful for precise comparison of related melodies, but often not well-suited for comparing across cultures. Cantometrics is well-suited for broad cross-cultural comparison, but less well-suited for comparing fine-grained differences within cultures or very broad comparisons between human and animal vocalisations. Automatic analyses allow for more detailed and objective comparisons of music and speech, but struggle to capture some of the complexities of polyphonic music and extract features such as scale tunings from the imprecision found in real singing. And we haven’t even touched on more sophisticated automated methods (e.g., Music Information Retrieval Toolbox [Lartillot et al., 2008]; Essentia [Bogdanov et al., 2013]; Librosa [McFee et al., 2015])16 or other comparative methods such as qualitative analysis or approaches based on Indigenous knowledge (for a very brief glimpse of those, see the Epilogue).
Overview of research methods: pros and cons.