Skip to main content

Structural sparsification for far-field speaker recognition with intel R GNA

Publication ,  Conference
Zhang, J; Huang, J; Deisher, M; Li, H; Chen, Y
Published in: ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
May 1, 2020

Recently, deep neural networks (DNN) have been widely used in speaker recognition area. In order to achieve fast response time and high accuracy, the requirements for hardware resources increase rapidly. However, as the speaker recognition application is often implemented on mobile devices, it is necessary to maintain a low computational cost while keeping high accuracy in far-field condition. In this paper, we apply structural sparsification on time-delay neural networks (TDNN) to remove redundant structures and accelerate the execution. On our targeted hardware, our model can remove 60% of parameters and only slightly increasing equal error rate (EER) by 0.18% while our structural sparse model can achieve more than 1.5× speedup.

Duke Scholars

Published In

ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

DOI

ISSN

1520-6149

Publication Date

May 1, 2020

Volume

2020-May

Start / End Page

3037 / 3041
 

Citation

APA
Chicago
ICMJE
MLA
NLM
Zhang, J., Huang, J., Deisher, M., Li, H., & Chen, Y. (2020). Structural sparsification for far-field speaker recognition with intel R GNA. In ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings (Vol. 2020-May, pp. 3037–3041). https://doi.org/10.1109/ICASSP40776.2020.9054569
Zhang, J., J. Huang, M. Deisher, H. Li, and Y. Chen. “Structural sparsification for far-field speaker recognition with intel R GNA.” In ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, 2020-May:3037–41, 2020. https://doi.org/10.1109/ICASSP40776.2020.9054569.
Zhang J, Huang J, Deisher M, Li H, Chen Y. Structural sparsification for far-field speaker recognition with intel R GNA. In: ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. 2020. p. 3037–41.
Zhang, J., et al. “Structural sparsification for far-field speaker recognition with intel R GNA.” ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, vol. 2020-May, 2020, pp. 3037–41. Scopus, doi:10.1109/ICASSP40776.2020.9054569.
Zhang J, Huang J, Deisher M, Li H, Chen Y. Structural sparsification for far-field speaker recognition with intel R GNA. ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. 2020. p. 3037–3041.

Published In

ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

DOI

ISSN

1520-6149

Publication Date

May 1, 2020

Volume

2020-May

Start / End Page

3037 / 3041