Variational Bayesian Methods for Audio Indexing

Valente, Fabio; Wellekens, C.

doi:10.1007/11677482_27

Cited by 22 publications

(32 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…It enables an inference problem to be converted to an optimisation problem by approximating the intractable distribution with a tractable approximation obtained by minimising the Kullback-Leibler divergence between them. In [23] a Variational Bayes-EM algorithm is used to learn a GMM speaker model and optimize a change detection process and the merging criterion. In [24] Variational Bayes is combined successfully with eigenvoice modeling, described in [25], for the speaker diarization of telephone conversations.…”

Section: ) Bottom-up Approachmentioning

confidence: 99%

Speaker Diarization: A Review of Recent Research

Anguera

Bozonnet²,

Evans³

et al. 2012

IEEE Trans. Audio Speech Lang. Process.

571

349

View full text Add to dashboard Cite

Abstract-Speaker diarization is the task of determining "who spoke when?" in an audio or video recording that contains an unknown amount of speech and also an unknown number of speakers. Initially, it was proposed as a research topic related to automatic speech recognition, where speaker diarization serves as an upstream processing step. Over recent years, however, speaker diarization has become an important key technology for many tasks, such as navigation, retrieval, or higher-level inference on audio data. Accordingly, many important improvements in accuracy and robustness have been reported in journals and conferences in the area. The application domains, from broadcast news, to lectures and meetings, vary greatly and pose different problems, such as having access to multiple microphones and multimodal information or overlapping speech. The most recent review of existing technology dates back to 2006 and focuses on the broadcast news domain. In this paper we review the current state-of-the-art, focusing on research developed since 2006 that relates predominantly to speaker diarization for conference meetings. Finally, we present an analysis of speaker diarization performance as reported through the NIST Rich Transcription evaluations on meeting data and identify important areas for future research.

show abstract

Section: ) Bottom-up Approachmentioning

confidence: 99%

Speaker Diarization: A Review of Recent Research

Anguera

Bozonnet²,

Evans³

et al. 2012

IEEE Trans. Audio Speech Lang. Process.

571

349

View full text Add to dashboard Cite

show abstract

“…There exists a large amount of previous work on the diarization problem, much of which is reviewed in [1]- [3]. Because of its relative simplicity, the Bayesian Information Criterion (BIC) has served as a backbone and an inspiration for the development of a number of initial approaches involving speaker change detection and bottom-up hierarchical clustering [4], [5]. Bottom-up approaches in general, where a number of clusters or models are trained and successively merged until only one remains for each speaker, are easily the most popular in the community and consistently tend to achieve the state-of-the-art [6], [7].…”

Section: Introductionmentioning

confidence: 99%

Unsupervised Methods for Speaker Diarization: An Integrated and Iterative Approach

Shum

Dehak

et al. 2013

IEEE Trans. Audio Speech Lang. Process.

166

View full text Add to dashboard Cite

Abstract-In speaker diarization, standard approaches typically perform speaker clustering on some initial segmentation before refining the segment boundaries in a re-segmentation step to obtain a final diarization hypothesis. In this paper, we integrate an improved clustering method with an existing re-segmentation algorithm and, in iterative fashion, optimize both speaker cluster assignments and segmentation boundaries jointly. For clustering, we extend our previous research using factor analysis for speaker modeling. In continuing to take advantage of the effectiveness of factor analysis as a front-end for extracting speaker-specific features (i.e., i-vectors), we develop a probabilistic approach to speaker clustering by applying a Bayesian Gaussian Mixture Model (GMM) to principal component analysis (PCA)-processed i-vectors. We then utilize information at different temporal resolutions to arrive at an iterative optimization scheme that, in alternating between clustering and re-segmentation steps, demonstrates the ability to improve both speaker cluster assignments and segmentation boundaries in an unsupervised manner. Our proposed methods attain results that are comparable to those of a state-of-the-art benchmark set on the multi-speaker CallHome telephone corpus. We further compare our system with a Bayesian nonparametric approach to diarization and attempt to reconcile their differences in both methodology and performance.

show abstract

“…marginal likelihoods, posterior probabilities, predictive densities) by bounding the marginal likelihood of the model from below. The use of VB is SD has been pioneered by F. Valente (Valente, 2005) and has been refined by P. Kenny et al (Kenny et al, 2010) by applying it to i-vectors. We should emphasize that VB is a general purpose (approximate) inference method and its use is not limited to finite mixture models.…”

Section: Methods Based On Variational Bayes Approximate Inferencementioning

confidence: 99%

“…Each supervector is then projected onto a space of lower dimensionality and VB inference in adopted to estimate the number of speakers and the assignment of segments to speakers. VB methods that do not make use of the supervector representation can be found in (Valente, 2005),…”

Section: A Variational Bayes Approach To Speaker Diarization Using Sumentioning

confidence: 99%

“…on the hypothesis), and select the order that maximizes the evidence (or an approximation) of the model. The Evolving-HMM and the VB approach of (Valente, 2005) are typical examples of this framework. However, such exhaustive search solutions may lack of efficiency, especially in cases where the hypothesis space is quite large (e.g.…”

Section: General About Infinite Modelsmentioning

confidence: 99%

See 1 more Smart Citation