Hiding Speaker's Sex in Speech Using Zero-Evidence Speaker Representation in an Analysis/Synthesis Pipeline
Publication
, Conference
Noe, PG; Miao, X; Wang, X; Yamagishi, J; Bonastre, JF; Matrouf, D
Published in: ICASSP IEEE International Conference on Acoustics Speech and Signal Processing Proceedings
January 1, 2023
The use of modern vocoders in an analysis/synthesis pipeline allows us to investigate high-quality voice conversion that can be used for privacy purposes. Here, we propose to transform the speaker embedding and the pitch in order to hide the sex of the speaker. ECAPA-TDNN-based speaker representation fed into a HiFiGAN vocoder is protected using a neural-discriminant analysis approach, which is consistent with the zero-evidence concept of privacy. This approach significantly reduces the information in speech related to the speaker's sex while preserving speech content and some consistency in the resulting protected voices.
Duke Scholars
Published In
ICASSP IEEE International Conference on Acoustics Speech and Signal Processing Proceedings
DOI
ISSN
1520-6149
Publication Date
January 1, 2023
Volume
2023-June
Citation
APA
Chicago
ICMJE
MLA
NLM
Noe, P. G., Miao, X., Wang, X., Yamagishi, J., Bonastre, J. F., & Matrouf, D. (2023). Hiding Speaker's Sex in Speech Using Zero-Evidence Speaker Representation in an Analysis/Synthesis Pipeline. In ICASSP IEEE International Conference on Acoustics Speech and Signal Processing Proceedings (Vol. 2023-June). https://doi.org/10.1109/ICASSP49357.2023.10096749
Noe, P. G., X. Miao, X. Wang, J. Yamagishi, J. F. Bonastre, and D. Matrouf. “Hiding Speaker's Sex in Speech Using Zero-Evidence Speaker Representation in an Analysis/Synthesis Pipeline.” In ICASSP IEEE International Conference on Acoustics Speech and Signal Processing Proceedings, Vol. 2023-June, 2023. https://doi.org/10.1109/ICASSP49357.2023.10096749.
Noe PG, Miao X, Wang X, Yamagishi J, Bonastre JF, Matrouf D. Hiding Speaker's Sex in Speech Using Zero-Evidence Speaker Representation in an Analysis/Synthesis Pipeline. In: ICASSP IEEE International Conference on Acoustics Speech and Signal Processing Proceedings. 2023.
Noe, P. G., et al. “Hiding Speaker's Sex in Speech Using Zero-Evidence Speaker Representation in an Analysis/Synthesis Pipeline.” ICASSP IEEE International Conference on Acoustics Speech and Signal Processing Proceedings, vol. 2023-June, 2023. Scopus, doi:10.1109/ICASSP49357.2023.10096749.
Noe PG, Miao X, Wang X, Yamagishi J, Bonastre JF, Matrouf D. Hiding Speaker's Sex in Speech Using Zero-Evidence Speaker Representation in an Analysis/Synthesis Pipeline. ICASSP IEEE International Conference on Acoustics Speech and Signal Processing Proceedings. 2023.
Published In
ICASSP IEEE International Conference on Acoustics Speech and Signal Processing Proceedings
DOI
ISSN
1520-6149
Publication Date
January 1, 2023
Volume
2023-June