An Attribute-Aligned Strategy for Learning Speech Representation
Published in ISCA Interspeech, 2021
Recommended citation: Huang, Y. L., Su, B. H., Hong, Y. W. P., & Lee, C. C. (2021). "An Attribute-Aligned Strategy for Learning Speech Representation." arXiv preprint arXiv:2106.02810. 1(1). https://www.isca-speech.org/archive/pdfs/interspeech_2021/huang21b_interspeech.pdf
To protect the privacy issue of end-users, we design a flexible attribute align strategy for distributing the information according to the pre-defined order. The overall architecture is implemented integrating the concept of dropout function and achieving the competitive results while considering the privacy compared to the SOTA.
Recommended citation: Huang, Y. L., Su, B. H., Hong, Y. W. P., & Lee, C. C. (2021). “An Attribute-Aligned Strategy for Learning Speech Representation.” arXiv preprint arXiv:2106.02810. 1(1).