An Analysis of Low-Arousal Piano Music Ratings to Uncover What Makes Calm and Sad Music So Difficult to Distinguish in Music Emotion Recognition

Hong Y, Chau CJ, Horner A. An Analysis of Low-Arousal Piano Music Ratings to Uncover What Makes Calm and Sad Music So Difficult to Distinguish in Music Emotion Recognition. J Audio Eng Soc. 2017;65(4):304-320. doi:10.17743/jaes.2017.0001

@article{Hong2017_18563,
  author = {Hong, Yu and Chau, Chuck-Jee and Horner, Andrew},
  title = {{An Analysis of Low-Arousal Piano Music Ratings to Uncover What Makes Calm and Sad Music So Difficult to Distinguish in Music Emotion Recognition}},
  journal = {Journal of the Audio Engineering Society},
  volume = {65},
  number = {4},
  pages = {304--320},
  year = {2017},
  month = apr,
  publisher = {Audio Engineering Society},
  doi = {10.17743/jaes.2017.0001},
  url = {https://doi.org/10.17743/jaes.2017.0001}
}

TY  - JOUR
TI  - An Analysis of Low-Arousal Piano Music Ratings to Uncover What Makes Calm and Sad Music So Difficult to Distinguish in Music Emotion Recognition
AU  - Hong, Yu
AU  - Chau, Chuck-Jee
AU  - Horner, Andrew
T2  - Journal of the Audio Engineering Society
J2  - J. Audio Eng. Soc.
VL  - 65
IS  - 4
SP  - 304
EP  - 320
PY  - 2017
DA  - 2017/04/06
DO  - 10.17743/jaes.2017.0001
UR  - https://doi.org/10.17743/jaes.2017.0001
PB  - Audio Engineering Society
LA  - en
AB  - Systems that recognize the emotional content of music and systems that provide music recommendations often use a simplified 4-quadrant model with categories such as Happy, Sad, Angry, and Calm. Previous research has shown that both listeners and automated systems often have difficulty distinguishing low-arousal categories such as Calm and Sad. This paper explores what makes these categories difficult to distinguish. 300 low-arousal excerpts from the classical piano repertoire were used to determine the coverage of the categories Calm and Sad in the low-arousal space, their overlap, and their balance to one another. Results show that Calm was 40% bigger in terms of coverage than Sad, but on average, Sad excerpts were significantly more negative in mood than Calm excerpts that were positive. Calm and Sad overlapped in nearly 20% of the excerpts, meaning 20% of the excerpts were about equally Calm and Sad. Calm and Sad covered about 92% of the low-arousal space. The largest holes were for excerpts considered Mysterious and Doubtful. Due to the holes in the coverage, the overlaps, and imbalances, the Calm-Sad model adds about 6% more errors when compared to asking users directly whether the mood of the music is positive or negative. Nevertheless, the Calm-Sad model is still useful and appropriate for many applications.
ER  -