About the FAV Corpus (link below)
This corpus, which we call the FAV Corpus, contains audio files of music provided by 140 respondents to a survey, wherein we asked each respondent to provide three of their favorite pieces along with their favorite 15-second excerpt from each piece.
The “excerptAudio” folder contains audio files of the 15-second excerpts.
The “musAudio" folders contain audio files of the complete pieces.
The “merged_data” spreadsheet indicates, for each item, (a) the respondent’s number (respondents were assigned numbers arbitrarily), (b) the excerpt number for that respondent (1, 2, or 3), (c) the artist and title of the piece, (d) the style and historical era or year, (e) the duration of the piece, (f) the timepoints of the preferred excerpt, (g) whether the respondent indicated that they enjoyed the excerpt “much more than” [A] or “about as much as” [B] the rest of the piece, and (h) the respondent’s comment about why they liked the excerpt.
The “Formal Analyses” folder contains text files with analyses of the musical form of a subset of the pieces in the corpus. That folder contains a “README” file which describes the structure of the analysis files and the methodology of analysis.
For further details on the FAV Corpus, the survey methodology, and analysis of the corpus, see Ethan Lustig’s Ph.D. dissertation, "The effect of perceived complexity and formal location on musical preference" (2021, Eastman School of Music, University of Rochester) and our forthcoming ISMIR paper, "The FAV Corpus: An audio dataset of favorite pieces and excerpts, with formal analyses and music theory descriptors" (2023).
-Ethan Lustig & David Temperley
The “excerptAudio” folder contains audio files of the 15-second excerpts.
The “musAudio" folders contain audio files of the complete pieces.
The “merged_data” spreadsheet indicates, for each item, (a) the respondent’s number (respondents were assigned numbers arbitrarily), (b) the excerpt number for that respondent (1, 2, or 3), (c) the artist and title of the piece, (d) the style and historical era or year, (e) the duration of the piece, (f) the timepoints of the preferred excerpt, (g) whether the respondent indicated that they enjoyed the excerpt “much more than” [A] or “about as much as” [B] the rest of the piece, and (h) the respondent’s comment about why they liked the excerpt.
The “Formal Analyses” folder contains text files with analyses of the musical form of a subset of the pieces in the corpus. That folder contains a “README” file which describes the structure of the analysis files and the methodology of analysis.
For further details on the FAV Corpus, the survey methodology, and analysis of the corpus, see Ethan Lustig’s Ph.D. dissertation, "The effect of perceived complexity and formal location on musical preference" (2021, Eastman School of Music, University of Rochester) and our forthcoming ISMIR paper, "The FAV Corpus: An audio dataset of favorite pieces and excerpts, with formal analyses and music theory descriptors" (2023).
-Ethan Lustig & David Temperley
