Fast adaptive component weighted cepstrum pole filtering for speaker identification

Swanson, A.L.; Ramachandran, Ravi P.; Chin, Steven H.

doi:10.1109/iscas.2004.1329882

Cited by 4 publications

(3 citation statements)

References 13 publications

(15 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For each selected frame, seven 12 dimensional features are calculated: (1) linear predictive cepstrum (CEP), (2) adaptive component weighted (ACW) cepstrum [64], (3) postfilter (PFL) cepstrum [64], (4) pole filtered mean removed cepstrum (PFMRCEP) [67], (5) mean removed ACW cepstrum (MRACW), (6) pole filtered mean removed ACW cepstrum (PFMRACW) [68] and (7) mean removed PFL cepstrum (MRPFL). Seven different feature vectors of dimension 12 are computed.…”

Section: A Speaker Identification Projectmentioning

confidence: 99%

Vertical Integration of Biometrics Across the Curriculum: Case Study of Speaker, Face and Iris Recognition

Ramachandran

Dahm

Nickel

et al. 2014

IEEE Circuits Syst. Mag.

View full text Add to dashboard Cite

Vertical integration is a powerful curricular tool that allows students to better appreciate the interconnections among the concepts acquired and learned in different courses. it can be used to bring a modern topic at all levels of the undergraduate curriculum with little additional resources. this paper gives a brief survey of various vertical integration efforts and describes one effort at integrating biometrics throughout the curriculum. the focus is on three senior level projects (speaker, face and iris recognition) that not only rely on vertical integration but also reinforce design, software skills and knowledge of stEm concepts. the freshman through junior levels are also described. the assessment results show that students acquire specific learning outcomes and perceive the value of vertical integration. iEEE circUits And systEms mAgAzinE third QUArtEr 2014

show abstract

Section: A Speaker Identification Projectmentioning

confidence: 99%

Vertical Integration of Biometrics Across the Curriculum: Case Study of Speaker, Face and Iris Recognition

Ramachandran

Dahm

Nickel

et al. 2014

IEEE Circuits Syst. Mag.

View full text Add to dashboard Cite

show abstract

“…The PFMRACW is a feature which combines Adaptive Component Weighted (ACW) cepstrum and pole filtering [6], [7] thereby leading to enhanced robustness to channel effects. The pole filtered ACW transfer function can be written as the ratio of two transfer functions M (z) and A(z/γ) as shown in Eq.…”

Section: Pole Filtered Mean Removed Adaptive Component Weighted Cementioning

confidence: 99%

“…The PFMRCEP feature generally provides a large ISR when observed on sessions 2-5. However, the ISR on session [6][7][8][9][10] show that the PFMRACW can be used to provide a higher 2010 Asia Pacific Conference on Circuits and Systems (APCCAS 2010) 6-9 December 2010, Kuala Lumpur, Malaysia ISR than the PFMRCEP. This indicates that the PFMRACW is more robust to changes in the channel than the PFMRCEP.…”

Section: B Proceduresmentioning

confidence: 99%

Fusion methods for boosting performance of speaker identification systems

Ditzler

Ethridge

Ramachandran

et al. 2010

2010 IEEE Asia Pacific Conference on Circuits and Systems

View full text Add to dashboard Cite

Abstract-Two important components of a speaker identification system are the feature extraction and the classification tasks. First, features must be robust to noise and they must also be able to provide discriminating information that the classifier can use to determine the speaker's identity. Second, the classifier must take the features that have been extracted from a sentence and label them as corresponding to one of the enrolled speakers. However, sets of features may be even more beneficial than any single feature by itself. There may be information present in one feature that other features do not have. Therefore, we present analysis of features and fusion by employing probabilistic averaging and weighted majority voting. Weighted voting will require that the weights are determined in a non-heuristic methodology and are robust to data with a large amount of channel distortion. Results using the King database show that both fusion methods lead to enhanced performance.

show abstract