TY - GEN
T1 - Adaptation of PLDA for multi-source text-independent speaker verification
AU - Chen, Liping
AU - Lee, Kong Aik
AU - Ma, Bin
AU - Ma, Long
AU - Li, Haizhou
AU - Dai, Li Rong
N1 - Publisher Copyright:
© 2017 IEEE.
PY - 2017/6/16
Y1 - 2017/6/16
N2 - Probabilistic linear discriminant analysis (PLDA) is widely described as an effective model for text-independent speaker verification in the i-vector space. The PLDA scoring function is typically formulated as the likelihood ratio between the speaker-adapted and the universal PLDAs. In this case, the adaptation of PLDA was performed through the speaker factors. In this paper, we show that the channel factors of the PLDA could be equivalently exploited to deal with the multi-source conditions. In speaker verification, with the proposed method, a PLDAmodel trained on conversational telephone speech could be adequately adapted for interview-style microphone recordings. Experimental results on NIST SRE'08 and SRE'10 datasets confirm that the proposed method is effective, especially for the case whereby enrollment and test utterances were captured from different sources.
AB - Probabilistic linear discriminant analysis (PLDA) is widely described as an effective model for text-independent speaker verification in the i-vector space. The PLDA scoring function is typically formulated as the likelihood ratio between the speaker-adapted and the universal PLDAs. In this case, the adaptation of PLDA was performed through the speaker factors. In this paper, we show that the channel factors of the PLDA could be equivalently exploited to deal with the multi-source conditions. In speaker verification, with the proposed method, a PLDAmodel trained on conversational telephone speech could be adequately adapted for interview-style microphone recordings. Experimental results on NIST SRE'08 and SRE'10 datasets confirm that the proposed method is effective, especially for the case whereby enrollment and test utterances were captured from different sources.
KW - channel adaptation
KW - channel prior estimation
KW - multi-source speaker verification
KW - probabilistic linear discriminant analysis
UR - http://www.scopus.com/inward/record.url?scp=85023746705&partnerID=8YFLogxK
U2 - 10.1109/ICASSP.2017.7953184
DO - 10.1109/ICASSP.2017.7953184
M3 - Conference article published in proceeding or book
AN - SCOPUS:85023746705
T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
SP - 5380
EP - 5384
BT - 2017 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2017 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2017 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2017
Y2 - 5 March 2017 through 9 March 2017
ER -