Within-class covariance normalization for SVM-based speaker recognition

By Andrew O. Hatch, Sachin S. Kajarekar, Andreas Stolcke. Interspeech, 2006

466 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Signal Processing

View Paper: https://www.semanticscholar.org/paper/b0dd904d57d1629e747a32d41106611a8a2497f4

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Signal Processing

Citations in the corpus

Spoofing and countermeasures for speaker verification: A survey (2015, background): “, 2007) and within-class covariance normalisation (WCCN) (Hatch et al., 2006).”
Support vector machines versus fast scoring in the low-dimensional total variability space for speaker verification (2009, background): “Within class covariance normalization is presented in detail in [6] and is successfully applied in speaker factor space [4].”
Front-End Factor Analysis for Speaker Verification (2011, methodology): “mize the error rate, the author in [8] defines a set of upper bounds”
Probabilistic linear discriminant analysis of i-vector posterior distributions (2013, methodology): “A pre–processing step, which involves i–vector whitening followed by length normalization [10, 7], is required to achieve state–of–the– art results using i–vectors with Gaussian PLDA models.”
The Multiverse Loss for Robust Transfer Learning (2015, methodology): “It resembles, other methods that down-regulate the contribution of the directions in the vector space that account for much of the within class covariance, such as WCCN [9].”
ASVspoof 2019: A large-scale public database of synthesized, converted and replayed speech (2019, methodology): “ted for each speaker, and the whitened x-vectors were further normalized to have a unit length. After that, the processed x-vectors were transformed using within-class covariance normalization (WCCN) [50], where the 19 attacking systems and the bona ﬁde data were treated as 20 dierent classes. Final”
Speaker verification and spoken language identification using a generalized i-vector framework with phonetic tokenizations and tandem features (2014, methodology): “Third, within this i-vector space, variability compensation methods, such as Within-Class Covariance Nor-malization (WCCN) [4], Linear Discriminative Analysis (LDA) and Nuisance Attribute Projection (NAP) [5], are performed to reduce the variability for the subsequent modeling methods (e.g.,…”
Comparison of Speaker Recognition Approaches for Real Applications (2011, methodology): “These speaker features are finally normalized by means of Within Class Covariance Normalization (WCCN) [5][13].”
Generalized I-vector Representation with Phonetic Tokenizations and Tandem Features for both Text Independent and Text Dependent Speaker Verification (2015, methodology): “Third, within this i-vector space, variability compensation methods, such as Within-Class Co-variance Normalization (WCCN) [7], Linear Discrimi-native Analysis (LDA) and Nuisance Attribute Projection (NAP) [1] are performed to reduce the variability for subsequent scoring methods (e.g., cosine simil”
Language Recognition via i-vectors and Dimensionality Reduction (2011, background): “Every approach defines a new basis that maximizes the discrimination between the different language classes based on the defined criterion.”