Chapter 14 Speaker Recognition.ppt

1、Chapter 14 Speaker Recognition,14.1 Introduction to speaker recognition 14.2 The basic problems for speaker recognition 14.3 Approaches and systems 14.4 Language Identification,14.1 Introduction to speaker recognition (1),Speaker recognition tries to extract the personal factors of speakers, and spe

2、ech recognition common factors of speakers. Speaker Verification : is the speaker as claimed or not Speaker Identification : who is the speaker in a name list,Introduction to speaker recognition (2),Text dependent speaker recognition and text independent speaker recognition Application of speaker re

3、cognition Business Systems Legal Systems Military Systems Security Systems Hard problems Which features are effective and reliable ?,14.2 The basic problems of speaker recognition (1),System diagram Training : Create utterances and model parameters for all speakers Verification : Compare the real pa

4、rameters with that of claimed speaker. If the difference is less than some threshold the speaker is verified otherwise rejected,The basic problems of speaker recognition (2),Recognition : Compare the extracted parameters with the reference parameters of all speakers and chose the minimum distance de

5、cision to identify the speaker. Three basic problems : parameter selection; specifying the similarity measure to make the calculation simple and reliable; updating the reference parameters to adapt to users.,The basic problems of speaker recognition (3),Design compromising. For speaker verification

6、system, two important parameters are False Rejection Rate(FR) and False Acceptance Rate(FA). They have some relation with acceptance threshold. In different cases they have different influence. Performance vs number of speakers,The basic problems of speaker recognition (4),Updating of Reference Temp

7、lates Performance Evaluation Basic Characteristics of Speakers Ideally these features should effectively distinguish different speakers and keep relative stability when speech changes; they should be easy to extract; not easy to mimic.,The basic problems of speaker recognition (5),Evaluation approac

8、hes to parameters effectiveness F = /k,i. where xk(i)is the parameters for k-th utterance of i-th speaker. i is averaging for speakers; k is averaging for different utterances of a speaker; i=k is the mean estimation of i-th speaker; =i For multi-dimension, B=i, W=k,i. Divergence D = i,j,=Tr(W-1B),T

9、he basic problems of speaker recognition (6),Feature Examples (1) LPC and its derived parameters (2) Parameters deducted from speech spectrum (3) Mixture parameters Approaches to Speaker Recognition (1) Template Matching Method,The basic problems of speaker recognition (7),(2) Probability Model Meth

10、od (3) Text independent speaker recognition system based on VQ (4) Neural Network We have done a speaker recognition system using BP network. It was text dependent.,14.3 Approaches and systems (1),GMM(Gaussian Mixture Model) It is a kind of probability model. Every speaker corresponds a GMM model. P

11、(x|)=Pibi(x), i=1M It means P(x|) is the weighted sum of M normal density bi. x is n-dimensional observation vector; Pi are the weighting coefficients; bi are n-dimensional gaussian functions.,Approaches and systems (2),bi(x) = 1/(2)n/2|Ci|1/2* exp-(x-i)tCi-1(x-i)/2 where i is the mean vector, Ci is

12、 the covariance matrix. i =Pi, i, Ci, i=1M MLE of GMM parameters Assume X = xt t=1T are training feature vectors. The likelyhood of model is P(X|)=P(xt|), t=1T,Approaches and systems (3),The goal for training is to find a 0 such that P(X|0) get maximum: 0 =argmax P(X|) for all P(X|) is non linear fu

13、nction of . So EM is used to find the optimal . Define Q(,)=P(X,i|)log P(X,i|), i=1M, i is the sequence number of gaussian components. Q(,)= t(i)logPibi(xt), i=1M,t=1T,Approaches and systems (4),where t(i)= P(x|)P(it=i|xt,) P(it=i|xt,)=Pibi(xt)/Pmbm(xt) for m=1M Let the partial derivation of Q over

14、Pi,i,Ci, i=1M the following iteration could be : Pi=(1/T) P(it=i|xt,) i= P(it=i|xt,)xt / P(it=i|xt,), i=1M,And i2= P(it=i|xt,)xt2/ P(it=i|xt,), i=1MRecognition Algorithm As soon as we have models for all speakers then we will calculate maximum posteriori probability to find the speaker: S =argmax lo

15、gP(xt|k), t=1T, k=1M,Approaches and systems (5),14.4 Language Identification (1),Principles Different level to recognize and utilize : phonemes, syllable structure, prosodic features, lexicon categories, syntax and semantic network. Structure of Language Identification System Different systems : HMM

16、 based, phoneme based.,Language Identification (2),Our experimental system OGI corpus System now contains four languages : English, Chinese, Spanish, Japanese It is phoneme based. Every language has a set of HMM models for its phonemes. The models are constructed such that every phoneme could be fol

17、lowed by any phoneme. These models are trained by the corpus with the label files.,Language Identification (3),The system structure is similar with above. Every language has a network of HMM models. The incoming utterance (with different length) is feed into every language network and every network

18、will output a probability value for the incoming utterance. By comparing these values the decision could be made, the language will be identified. By our experiments the accuracy could be more than 95%. Because only thing is to decide,Language Identification (4),(continued) which language it is. So if you use two or three utterances to test, you will definitely get a correct answer with very high probability. The way is simpler than the large vocabulary word system based on HMM.,