Corpus ID: 7241984

Hunting for Wolves in Speaker Recognition

  title={Hunting for Wolves in Speaker Recognition},
  author={Lara Stoll and G. Doddington},
Identification and selection of speaker pairs that are difficult to distinguish offers the possibility of better focusing speaker recognition research, while also reducing the amount of data needed to estimate system performance with confidence. This work aims to predict which speaker pairs will be difficult for automatic speaker recognition systems to distinguish, by using features that characterize speakers, and thus provide a measure of speaker similarity. Features tested include pitch… Expand
Investigation of the effect of data duration and speaker gender on text-independent speaker recognition
It is shown that recognition accuracy for male speakers are higher than female independent from database and classifier and GSV-SVM system yields higher EER than other methods in the case of using short utterances. Expand
Towards Goat Detection in Text-Dependent Speaker Verification
A method that identifies speakers that are likely to have a high false-reject rate in a text-dependent speaker verification system (“goats”) is presented and it is shown that the 1% lowest-ranking enrollments have a false reject rate of up to 7.8%, compared to the system’s overall rate. Expand
Spoofing and countermeasures for speaker verification: A survey
A survey of past work and priority research directions for the future is provided, showing that future research should address the lack of standard datasets and the over-fitting of existing countermeasures to specific, known spoofing attacks. Expand
A Survey in Presentation Attack and Presentation Attack Detection
A general taxonomy of presentation attacks is proposed to cover different biometric modalities considering the attacker’s intention and the presentation instrument and mechanisms that aim to eliminate or mitigate those attacks are also taxonomized. Expand
Parole de locuteur : performance et confiance en identification biométrique vocale. (Speaker in speech : performance and confidence in voice biometric identification)
  • J. Kahn
  • Computer Science, Philosophy
  • 2011
Ce travail de these explore l’usage biometrique de la parole dont les applications sont tres nombreuses (securite, environnements intelligents, criminalistique, surveillance du territoire ouExpand


A na ve de-lambing method for speaker identification
A naive de-lambing method is described to deal with lamb speakers who are particularly easy to imitate in the closeset text-independent speaker identification system so as to improve the system performance. Expand
Modeling prosodic dynamics for speaker recognition
Two approaches that use the fundamental frequency and energy trajectories to capture long-term information are proposed that can achieve a 77% relative improvement over a system based on short-term pitch and energy features alone. Expand
Objectively measured descriptors applied to speaker characterization
A variety of features objectively extracted from the speech waveform to characterize perceptual speaker differences belong to three broad classes of glottal, vocal tract and prosodic measurements. Expand
SRI's 2004 NIST speaker recognition evaluation system
  • S. Kajarekar, L. Ferrer, +4 authors Jing Zheng
  • Computer Science
  • Proceedings. (ICASSP '05). IEEE International Conference on Acoustics, Speech, and Signal Processing, 2005.
  • 2005
A system that uses discriminant features from cepstral coefficients, and systems that use discriminant models from word n-grams and syllable-based NERF n- grams together with a cEPstral baseline system are evaluated. Expand
Using prosodic and conversational features for high-performance speaker recognition: report from JHU WS'02
  • B. Peskin, Jirí Navrátil, +4 authors Bing Xiang
  • Computer Science
  • 2003 IEEE International Conference on Acoustics, Speech, and Signal Processing, 2003. Proceedings. (ICASSP '03).
  • 2003
A variety of modeling techniques were examined, such as n-gram models of turn-level prosodic features and simple vectors of summary statistics per conversation side scored by k/sup th/ nearest-neighbor classifiers, finding that purely prosodic models were able to achieve equal error rates, and yielded significant gains when combined with more traditional systems. Expand
The NIST speaker recognition evaluation - Overview, methodology, systems, results, perspective
The performance trade-off of missed detections and false alarms for each system and the effects on performance of training condition, test segment duration, the speakers' sex and the match or mismatch of training and test handsets are presented. Expand
SHEEP, GOATS, LAMBS and WOLVES: a statistical analysis of speaker performance in the NIST 1998 speaker recognition evaluation
This paper proposes statistical tests for the existence of sheep, goats, lambs and wolves and applies these tests to hunt for such animals using results from the 1998 NIST speaker recognition evaluation. Expand
A distance measure between GMMs based on the unscented transform and its application to speaker recognition
This paper proposes an accurate and efficiently computed approximation of the KL-divergence based on the unscented transform which is usually used to obtain a better alternative to the extended Kalman filter and experimental results indicate that the proposed approximations outperform previously suggested methods. Expand
A study of speaker-distinguishing properties of the formant dynamics of /uː/ is presented. Measurements at equidistant intervals along the F1 and F2 contours of /uː/ are compared with polynomialExpand
Acoustic characteristics of speaker individuality: Control and conversion
A survey of non-parametric methods for spectral segmental characteristics mapping between speakers, introducing some different types of spectral mapping methods that have evolved in relation to the speaker adaptation techniques being developed in speech recognition research. Expand