DocumentCode
554101
Title
Notice of Retraction
Speaker classification based on high dimension feature vector
Author
Yi Yang ; Hui Song ; Jia Liu
Author_Institution
Dept. of Electron. Eng., Tsinghua Univ., Beijing, China
Volume
2
fYear
2011
fDate
26-28 July 2011
Firstpage
891
Lastpage
894
Abstract
Notice of Retraction
After careful and considered review of the content of this paper by a duly constituted expert committee, this paper has been found to be in violation of IEEE´s Publication Principles.
We hereby retract the content of this paper. Reasonable effort should be made to remove all past references to this paper.
The presenting author of this paper has the option to appeal this decision by contacting TPII@ieee.org.
Audio index is an important part of NIST-RT-SD evaluation since 2003. Speaker Diarization is one kind of audio index technology which is marked by different speakers. One essential component of speaker diarization is speaker clustering which is always the pre-processing of speech recognition. The general method is to extract acoustic feature such as LPCC or MFCC and achieve some model such as HMM or GHMM by training these data. Another way is to treat these data as some vectors and choose the distance criterion between two or more classes. The best DER score of NIST-RT-SD evaluation is 8.51% at 2007. We proposed a new spatial feature vector mixed with traditional acoustic feature vector. The spatial feature vector is provided by distributed microphones random arranged during the conference environment. The high-dimension SVM algorithm is utilized to classify the testing mixed feature vector after the training step accomplished. The experiment results show that the mixed feature vector can improve the classifier´s precision under meeting scene.
After careful and considered review of the content of this paper by a duly constituted expert committee, this paper has been found to be in violation of IEEE´s Publication Principles.
We hereby retract the content of this paper. Reasonable effort should be made to remove all past references to this paper.
The presenting author of this paper has the option to appeal this decision by contacting TPII@ieee.org.
Audio index is an important part of NIST-RT-SD evaluation since 2003. Speaker Diarization is one kind of audio index technology which is marked by different speakers. One essential component of speaker diarization is speaker clustering which is always the pre-processing of speech recognition. The general method is to extract acoustic feature such as LPCC or MFCC and achieve some model such as HMM or GHMM by training these data. Another way is to treat these data as some vectors and choose the distance criterion between two or more classes. The best DER score of NIST-RT-SD evaluation is 8.51% at 2007. We proposed a new spatial feature vector mixed with traditional acoustic feature vector. The spatial feature vector is provided by distributed microphones random arranged during the conference environment. The high-dimension SVM algorithm is utilized to classify the testing mixed feature vector after the training step accomplished. The experiment results show that the mixed feature vector can improve the classifier´s precision under meeting scene.
Keywords
audio signal processing; hidden Markov models; pattern classification; pattern clustering; speaker recognition; support vector machines; LPCC feature; MFCC feature; Mel frequency cepstral coefficient; SVM algorithm; acoustic feature extraction; audio index technology; feature vector; hidden Markov models; speaker classification; speaker clustering; speaker diarization; speech recognition; support vector machines; Delay effects; Feature extraction; Mel frequency cepstral coefficient; Microphones; Speech; Support vector machine classification; Training; Mixed Feature Vector; NIST Evaluation; Speaker Classification; Speaker Diarization; high-dimension SVM;
fLanguage
English
Publisher
ieee
Conference_Titel
Natural Computation (ICNC), 2011 Seventh International Conference on
Conference_Location
Shanghai
ISSN
2157-9555
Print_ISBN
978-1-4244-9950-2
Type
conf
DOI
10.1109/ICNC.2011.6022284
Filename
6022284
Link To Document