Toward Mass-Customizing Up/Down Generic 3-D Sounds for Listeners: A Pilot Experiment Concerning Inter-Subject Variability

Au J, So R, Horner A. Toward Mass-Customizing Up/Down Generic 3-D Sounds for Listeners: A Pilot Experiment Concerning Inter-Subject Variability. In: AES Convention 130. Audio Engineering Society; 2011. Paper 8406. Available from: https://aes.org/publications/elibrary-page/?id=15873

@inproceedings{Au2011_15873,
  author = {Au, John and So, Richard and Horner, Andrew},
  title = {{Toward Mass-Customizing Up/Down Generic 3-D Sounds for Listeners: A Pilot Experiment Concerning Inter-Subject Variability}},
  booktitle = {AES Convention 130},
  note = {Paper 8406},
  year = {2011},
  month = may,
  publisher = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=15873}
}

TY  - CPAPER
TI  - Toward Mass-Customizing Up/Down Generic 3-D Sounds for Listeners: A Pilot Experiment Concerning Inter-Subject Variability
AU  - Au, John
AU  - So, Richard
AU  - Horner, Andrew
T2  - AES Convention 130
M1  - Paper 8406
PY  - 2011
DA  - 2011/05/06
UR  - https://aes.org/publications/elibrary-page/?id=15873
PB  - Audio Engineering Society
LA  - en
AB  - The ‘delay-and-add’ theory (Hebrank and Wright, 1974) was adopted, modified, and implemented to calculate 4068 matching scores between each of the 192 HRTFs and the dimensions of 12 pairs of ears for two incident sound directions (30 degrees up and 30 degrees down). These novel scores (0 to 1) quantified the closeness between a HRTF and the individualized HRTF of a listener in terms of embedded spectral cues for localizing up/down directional sounds. Five HRTFs with 0, 25, 50, 75, and 100th percentile average matching scores were selected for the two incident directions. These 10 HRTFs were used to produce 10 sound cues (five from 30 degrees up and five from 30 degrees down). Ten listeners participated in a sound localization experiment to localize the ten sound cues presented in random order and with four repetitions. Preliminary results indicated that matching scores could explain up to 22% of the inter-subject variations in localization errors. Potential applications of matching scores to select the most appropriate non-individualized HRTF set for a listener are discussed.
ER  -