Online speaker diarization using adapted i-vector transforms

Zhu, Weizhong; Pelecanos, Jason

doi:10.1109/icassp.2016.7472638

Cited by 45 publications

(39 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This includes R = ∞, which is equal to not using adaptation. We may notice that the adaptation process proposed in [19] can improve the final DER in all four cases, but the individual segmentation approaches have different optimal values of R.…”

Section: Resultsmentioning

confidence: 95%

“…The basic structure of the system is based on the i-vector approach which has recently become standard in speaker diarization [12,19]. The specific implementation largely follows the descriptions presented in our previous papers [8,16] and a diagram of the main steps can be seen in Fig.…”

Section: Offline Diarization Systemmentioning

confidence: 99%

“…As the clustering step, we employ the i-vector adaptation process proposed by Zhu and Pelecanos [19], which is given by…”

Section: Online Systemmentioning

confidence: 99%

“…This means that a proper initial segmentation again becomes important. This paper is in part inspired by the recent work of Zhu and Pelecanos [19], who have proposed an incremental adaptation process for online i-vector based speaker diarization. Their original paper focuses mainly on the clustering step and sidesteps the question of segmentation by utilizing oracle segmentation based on reference transcripts (although this has very recently been extended with ASR-based segmentation [3]).…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Experiments with Segmentation in an Online Speaker Diarization System

Kunešová

Zajíc

Radová

2017

Text, Speech, and Dialogue

View full text Add to dashboard Cite

Abstract. In offline speaker diarization systems, particularly those aimed at telephone speech, the accuracy of the initial segmentation of a conversation is often a secondary concern. Imprecise segment boundaries are typically corrected during resegmentation, which is performed as the final step of the diarization process. However, such resegmentation is generally not possible in online systems, where past decisions are usually unchangeable. In such situations, correct segmentation becomes critical. In this paper, we evaluate several different segmentation approaches in the context of online diarization by comparing the overall performance of an i-vector-based diarization system set to operate in a sequential manner.

show abstract

Section: Resultsmentioning

confidence: 95%

Section: Offline Diarization Systemmentioning

confidence: 99%

“…As the clustering step, we employ the i-vector adaptation process proposed by Zhu and Pelecanos [19], which is given by…”

Section: Online Systemmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Experiments with Segmentation in an Online Speaker Diarization System

Kunešová

Zajíc

Radová

2017

Text, Speech, and Dialogue

View full text Add to dashboard Cite

show abstract

“…The GMM(Gaussian mixture model) in i-vector model can be seen as the combination of many Gaussian models, and each Gaussian model represents one element of music. The i-vector feature has been widely used in speaker recognition [5][6][7][8]. In this paper, the frame level features are used to train a i-vector extractor to get the segment level representation, and the GMM based universal Background model (GMM-UBM) is utilized to train the i-vector extractor.…”

Section: Introductionmentioning

confidence: 99%

Multilingual i-Vector Based Statistical Modeling for Music Genre Classification

2017

View full text Add to dashboard Cite

For music signal processing, compared with the strategy which models each short-time frame independently, when the long-time features are considered, the time-series characteristics of the music signal can be better presented. As a typical kind of long-time modeling strategy, the identification vector (i-vector) uses statistical modeling to model the audio signal in the segment level. It can better capture the important elements of the music signal, and these important elements may benefit to the classification of music signal. In this paper, the ivector based statistical feature for music genre classification is explored. In addition to learn enough important elements for music signal, a new multilingual i-vector feature is proposed based on the multilingual model. The experimental results show that the multilingual i-vector based models can achieve better classification performances than conventional short-time modeling based methods.

show abstract