This paper proposes a speaker-phonetic i-vector modeling method for text-dependent speaker verification with random digit strings, in which enrollment and test utterances are not of the same phrase. The core of the proposed method is making use of digit alignment information in i-vector framework. By utilizing force alignment information, verification scores of the testing trials can be computed in the fixed-phrase situation, in which the compared speech segments between the enrollment and test utterances are of the same phonetic content. Specifically, utterances are segmented into digits, then a unique phonetically-constrained i-vector extractor is applied to obtain speaker and channel variability representation for every digit segment. Probabilistic linear discriminant analysis (PLDA) and s-norm are subsequently used for channel compensation and score normalization respectively. The final score is obtained by combing the digit scores, which are computed by scoring individual digit segments of the test utterance against the corresponding ones of the enrollment. Experimental results on the Part 3 of Robust Speaker Recognition (RSR2015) database demonstrate that the proposed approach significantly outperforms GMM-UBM by 52.3% and 53.5% relative in equal error rate (EER) for male and female respectively.
Shengyu YAO
Chinese Academy of Sciences,University of Chinese Academy of Sciences
Ruohua ZHOU
Chinese Academy of Sciences,University of Chinese Academy of Sciences
Pengyuan ZHANG
Chinese Academy of Sciences,University of Chinese Academy of Sciences
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Shengyu YAO, Ruohua ZHOU, Pengyuan ZHANG, "Speaker-Phonetic I-Vector Modeling for Text-Dependent Speaker Verification with Random Digit Strings" in IEICE TRANSACTIONS on Information,
vol. E102-D, no. 2, pp. 346-354, February 2019, doi: 10.1587/transinf.2018EDP7310.
Abstract: This paper proposes a speaker-phonetic i-vector modeling method for text-dependent speaker verification with random digit strings, in which enrollment and test utterances are not of the same phrase. The core of the proposed method is making use of digit alignment information in i-vector framework. By utilizing force alignment information, verification scores of the testing trials can be computed in the fixed-phrase situation, in which the compared speech segments between the enrollment and test utterances are of the same phonetic content. Specifically, utterances are segmented into digits, then a unique phonetically-constrained i-vector extractor is applied to obtain speaker and channel variability representation for every digit segment. Probabilistic linear discriminant analysis (PLDA) and s-norm are subsequently used for channel compensation and score normalization respectively. The final score is obtained by combing the digit scores, which are computed by scoring individual digit segments of the test utterance against the corresponding ones of the enrollment. Experimental results on the Part 3 of Robust Speaker Recognition (RSR2015) database demonstrate that the proposed approach significantly outperforms GMM-UBM by 52.3% and 53.5% relative in equal error rate (EER) for male and female respectively.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2018EDP7310/_p
Copy
@ARTICLE{e102-d_2_346,
author={Shengyu YAO, Ruohua ZHOU, Pengyuan ZHANG, },
journal={IEICE TRANSACTIONS on Information},
title={Speaker-Phonetic I-Vector Modeling for Text-Dependent Speaker Verification with Random Digit Strings},
year={2019},
volume={E102-D},
number={2},
pages={346-354},
abstract={This paper proposes a speaker-phonetic i-vector modeling method for text-dependent speaker verification with random digit strings, in which enrollment and test utterances are not of the same phrase. The core of the proposed method is making use of digit alignment information in i-vector framework. By utilizing force alignment information, verification scores of the testing trials can be computed in the fixed-phrase situation, in which the compared speech segments between the enrollment and test utterances are of the same phonetic content. Specifically, utterances are segmented into digits, then a unique phonetically-constrained i-vector extractor is applied to obtain speaker and channel variability representation for every digit segment. Probabilistic linear discriminant analysis (PLDA) and s-norm are subsequently used for channel compensation and score normalization respectively. The final score is obtained by combing the digit scores, which are computed by scoring individual digit segments of the test utterance against the corresponding ones of the enrollment. Experimental results on the Part 3 of Robust Speaker Recognition (RSR2015) database demonstrate that the proposed approach significantly outperforms GMM-UBM by 52.3% and 53.5% relative in equal error rate (EER) for male and female respectively.},
keywords={},
doi={10.1587/transinf.2018EDP7310},
ISSN={1745-1361},
month={February},}
Copy
TY - JOUR
TI - Speaker-Phonetic I-Vector Modeling for Text-Dependent Speaker Verification with Random Digit Strings
T2 - IEICE TRANSACTIONS on Information
SP - 346
EP - 354
AU - Shengyu YAO
AU - Ruohua ZHOU
AU - Pengyuan ZHANG
PY - 2019
DO - 10.1587/transinf.2018EDP7310
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E102-D
IS - 2
JA - IEICE TRANSACTIONS on Information
Y1 - February 2019
AB - This paper proposes a speaker-phonetic i-vector modeling method for text-dependent speaker verification with random digit strings, in which enrollment and test utterances are not of the same phrase. The core of the proposed method is making use of digit alignment information in i-vector framework. By utilizing force alignment information, verification scores of the testing trials can be computed in the fixed-phrase situation, in which the compared speech segments between the enrollment and test utterances are of the same phonetic content. Specifically, utterances are segmented into digits, then a unique phonetically-constrained i-vector extractor is applied to obtain speaker and channel variability representation for every digit segment. Probabilistic linear discriminant analysis (PLDA) and s-norm are subsequently used for channel compensation and score normalization respectively. The final score is obtained by combing the digit scores, which are computed by scoring individual digit segments of the test utterance against the corresponding ones of the enrollment. Experimental results on the Part 3 of Robust Speaker Recognition (RSR2015) database demonstrate that the proposed approach significantly outperforms GMM-UBM by 52.3% and 53.5% relative in equal error rate (EER) for male and female respectively.
ER -