Room Acoustic Adversarial Neural Network for Robust Sound Event Classification

Upadhyaya S, Buyens W, Desmet W, Karsmakers P. Room Acoustic Adversarial Neural Network for Robust Sound Event Classification. J Audio Eng Soc. 2024;72(11):754-766. doi:10.17743/jaes.2022.0174

@article{Upadhyaya2024_22777,
  author = {Upadhyaya, Sreenivasa and Buyens, Wim and Desmet, Wim and Karsmakers, Peter},
  title = {{Room Acoustic Adversarial Neural Network for Robust Sound Event Classification}},
  journal = {Journal of the Audio Engineering Society},
  volume = {72},
  number = {11},
  pages = {754--766},
  year = {2024},
  month = nov,
  publisher = {Audio Engineering Society},
  doi = {10.17743/jaes.2022.0174},
  url = {https://doi.org/10.17743/jaes.2022.0174}
}

TY  - JOUR
TI  - Room Acoustic Adversarial Neural Network for Robust Sound Event Classification
AU  - Upadhyaya, Sreenivasa
AU  - Buyens, Wim
AU  - Desmet, Wim
AU  - Karsmakers, Peter
T2  - Journal of the Audio Engineering Society
J2  - J. Audio Eng. Soc.
VL  - 72
IS  - 11
SP  - 754
EP  - 766
PY  - 2024
DA  - 2024/11/13
DO  - 10.17743/jaes.2022.0174
UR  - https://doi.org/10.17743/jaes.2022.0174
PB  - Audio Engineering Society
LA  - en
AB  - The variation in the acoustic condition of a room presents a major hurdle in the performance robustness of sound event classification. Room impulse response characterizes the way in which a sound wave is propagated from source to receiver and the overall perceptual quality and intelligibility of the recorded sound. This study presents the Room Acoustic Adversarial Neural Network (RAANN) method that can make sound event classification more robust to changes in acoustic condition by exploiting knowledge regarding the room acoustics during learning. With RAANN, the weighted F1 score for the classification task improved by 1.54 percentage points, and the standard deviation in performance dropped from 1.74 percentage points to 1.07 percentage points for acoustic conditions that were harder than those seen during the learning phase. The Clarity Index over the first 25 ms emerged as a good metric for the acoustic estimation in the RAANN training.
ER  -