Trained Listeners Assessing the xHE-AAC Codec Performance for Streaming Audio-for-Video Broadcast

Bielski L, Thompson C, Shaffer L, Sollars S, Bolin J, Williams P. Trained Listeners Assessing the xHE-AAC Codec Performance for Streaming Audio-for-Video Broadcast. In: AES Conference: AES 2023 International Conference on Audio Education. Audio Engineering Society; 2023. Paper 2. Available from: https://aes.org/publications/elibrary-page/?id=22228

@inproceedings{Bielski2023_22228,
  author = {Bielski, Lynn and Thompson, Christoph and Shaffer, Lauren and Sollars, Stan and Bolin, Jocelyn and Williams, Phillip},
  title = {{Trained Listeners Assessing the xHE-AAC Codec Performance for Streaming Audio-for-Video Broadcast}},
  booktitle = {AES Conference: AES 2023 International Conference on Audio Education},
  note = {Paper 2},
  year = {2023},
  month = sep,
  publisher = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=22228}
}

TY  - CPAPER
TI  - Trained Listeners Assessing the xHE-AAC Codec Performance for Streaming Audio-for-Video Broadcast
AU  - Bielski, Lynn
AU  - Thompson, Christoph
AU  - Shaffer, Lauren
AU  - Sollars, Stan
AU  - Bolin, Jocelyn
AU  - Williams, Phillip
T2  - AES Conference: AES 2023 International Conference on Audio Education
M1  - Paper 2
PY  - 2023
DA  - 2023/09/06
UR  - https://aes.org/publications/elibrary-page/?id=22228
PB  - Audio Engineering Society
LA  - en
AB  - This paper describes student participant training, content selection, and administration of a subjective test to evaluate xHE-AAC. The primary goal was to examine the possibility that students could be trained to perceive impairments in broadcast content. Years of musical training, ability to distinguish bitrates, and time required for rating the materials was examined. Data on how long listeners listen to the references, the coded signals, and the influence of these factors on ratings were analyzed. Listeners were trained to perceive impairments in broadcast content. A secondary goal was to find the bit rate at which the codec xHE-AAC becomes transparent. Eight audio excerpts from the Netflix catalogue were processed at 64, 128 and 192 kbps and were compared to 48 kHz, 32-bit floating reference samples, using the ITU-R BS.1116-3 protocol. Impairment scores were statistically different from the reference for six of eight audio samples at 64 kbps, and for three audio samples at 128 kbps. Mean impairment scores for 192 kbps were not significantly lower than scores of the reference for five samples. Listeners perceived artifacts in most audio samples at 64 kbps. At 128 kbps artifacts were detected in fewer samples, but with greater consistency than at 192 kbps. The results indicate student listeners can be trained to perceive and evaluate impairments in broadcast content.
ER  -