CONF Dey_ICASSP_2016/IDIAP DEEP NEURAL NETWORK BASED POSTERIORS FOR TEXT-DEPENDENT SPEAKER VERIFICATION Dey, Subhadeep Madikeri, Srikanth Ferras, Marc Motlicek, Petr EXTERNAL https://publications.idiap.ch/attachments/papers/2016/Dey_ICASSP_2016.pdf PUBLIC https://publications.idiap.ch/index.php/publications/showcite/Dey_Idiap-RR-08-2016 Related documents Proceedings of 2016 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2016) Shanghai 2016 IEEE 5050-5054 The i-vector and Joint Factor Analysis (JFA) systems for text- dependent speaker verification use sufficient statistics computed from a speech utterance to estimate speaker models. These statis- tics average the acoustic information over the utterance thereby losing all the sequence information. In this paper, we study ex- plicit content matching using Dynamic Time Warping (DTW) and present the best achievable error rates for speaker-dependent and speaker-independent content matching. For this purpose, a Deep Neural Network/Hidden Markov Model Automatic Speech Recog- nition (DNN/HMM ASR) system is used to extract content-related posterior probabilities. This approach outperforms systems using Gaussian mixture model posteriors by at least 50% Equal Error Rate (EER) on the RSR2015 in content mismatch trials. DNN posteriors are also used in i-vector and JFA systems, obtaining EERs as low as 0.02%. REPORT Dey_Idiap-RR-08-2016/IDIAP DEEP NEURAL NETWORK BASED POSTERIORS FOR TEXT-DEPENDENT SPEAKER VERIFICATION Dey, Subhadeep Madikeri, Srikanth Ferras, Marc Motlicek, Petr Idiap-RR-08-2016 2016 Idiap April 2016 The i-vector and Joint Factor Analysis (JFA) systems for text- dependent speaker verification use sufficient statistics computed from a speech utterance to estimate speaker models. These statis- tics average the acoustic information over the utterance thereby losing all the sequence information. In this paper, we study ex- plicit content matching using Dynamic Time Warping (DTW) and present the best achievable error rates for speaker-dependent and speaker-independent content matching. For this purpose, a Deep Neural Network/Hidden Markov Model Automatic Speech Recog- nition (DNN/HMM ASR) system is used to extract content-related posterior probabilities. This approach outperforms systems using Gaussian mixture model posteriors by at least 50% Equal Error Rate (EER) on the RSR2015 in content mismatch trials. DNN posteriors are also used in i-vector and JFA systems, obtaining EERs as low as 0.02%.