Efficient Audio Enhancement With a Differentiable Psychoacoustic Loss

Abreu W, Miranda BV, Biscainho LWP. Efficient Audio Enhancement With a Differentiable Psychoacoustic Loss. J Audio Eng Soc. 2026;74(6):430-444. doi:10.17743/jaes.2022.0277

@article{Abreu2026_23240,
  author = {Abreu, Wallace and Miranda, Bernardo V. and Biscainho, Luiz W. P.},
  title = {{Efficient Audio Enhancement With a Differentiable Psychoacoustic Loss}},
  journal = {Journal of the Audio Engineering Society},
  volume = {74},
  number = {6},
  pages = {430--444},
  year = {2026},
  month = jun,
  publisher = {Audio Engineering Society},
  doi = {10.17743/jaes.2022.0277},
  url = {https://doi.org/10.17743/jaes.2022.0277}
}

TY  - JOUR
TI  - Efficient Audio Enhancement With a Differentiable Psychoacoustic Loss
AU  - Abreu, Wallace
AU  - Miranda, Bernardo V.
AU  - Biscainho, Luiz W. P.
T2  - Journal of the Audio Engineering Society
J2  - J. Audio Eng. Soc.
VL  - 74
IS  - 6
SP  - 430
EP  - 444
PY  - 2026
DA  - 2026/06/08
DO  - 10.17743/jaes.2022.0277
UR  - https://doi.org/10.17743/jaes.2022.0277
PB  - Audio Engineering Society
LA  - en
AB  - Audio enhancement aims to improve the perceived quality of audio signals. Initially, to address bandwidth extension, this work proposes AEROMambaP, an efficient variant of the AERO super-resolution architecture in which attention and long short-term memory layers are replaced by the Mamba state-space model, together with a newly developed differentiable perceptual loss derived from the perceptual audio quality measure (PAQM). During training, the architecture requires two to fours times less GPU memory than the baseline; during inference, it achieves a 14× speedup while using one-fifth of the GPU memory. When upsampling both a piano dataset and MUSDB18 from 11.025 kHz to 44.1 kHz, subjective listening tests show that AEROMambaP outperforms AERO by 15% in perceived quality scores. Next, to enhance of audio signals highly compressed by lossy coding, it is further proposed AEROMambaPS̄, which applies the same framework but replaces short-time Fourier transform reconstruction losses with the PAQM loss, specifically to enhance MP3-encoded audio at 32 kbps. In listening evaluations, AEROMambaPS̄ achieves 52% higher quality rating than AEROMambaP when restoring compressed audio. These results demonstrate that PAQM-driven training coupled with lightweight state-space modeling yields high perceptual quality and computational efficiency in both band-limited and compressed audio scenarios.
ER  -