Validation of a Neural Network Clustering Model for Affective Response to Immersive Music

Kim S, Ko D, Howie W. Validation of a Neural Network Clustering Model for Affective Response to Immersive Music. In: AES Convention 155. Audio Engineering Society; 2023. Paper 162. Available from: https://aes.org/publications/elibrary-page/?id=22316

@inproceedings{Kim2023_22316,
  author = {Kim, Sungyoung and Ko, Doyuen and Howie, Will},
  title = {{Validation of a Neural Network Clustering Model for Affective Response to Immersive Music}},
  booktitle = {AES Convention 155},
  note = {Paper 162},
  year = {2023},
  month = oct,
  publisher = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=22316}
}

TY  - CPAPER
TI  - Validation of a Neural Network Clustering Model for Affective Response to Immersive Music
AU  - Kim, Sungyoung
AU  - Ko, Doyuen
AU  - Howie, Will
T2  - AES Convention 155
M1  - Paper 162
PY  - 2023
DA  - 2023/10/06
UR  - https://aes.org/publications/elibrary-page/?id=22316
PB  - Audio Engineering Society
LA  - en
AB  - Understanding differences between unique individuals is an important and emerging topic in auditory science and immersive experiences. Socio-cultural and anthropometric idiosyncrasies of listeners can lead to unintended auditory experiences, far from what media content creators intended. To better understand how this individuality may influence a listener’s preferences, we investigated various individually related factors, including previous listening experiences and cognitive profiles. In addition, we proposed a data-driven clustering method and showed its efficacy for meaningful grouping of listeners. In this study, we validated the data-driven method with 13 new subjects who generated attribute rating data for 16 stimulus conditions. The method, employing neural network clustering, successfully grouped participants into two preference-based categories with a 92.3% accuracy. The results support the proposed model’s reliability and its potential in applications to enhance individually optimized 3D music presentations.
ER  -