Sorry, you need to enable JavaScript to visit this website.

facebooktwittermailshare

Speaker Diarisation Using 2D Self-attentive Combination of Embeddings

Abstract: 

Speaker diarisation systems often cluster audio segments using speaker embeddings such as i-vectors and d-vectors. Since different types of embeddings are often complementary, this paper proposes a generic framework to improve performance by combining them into a single embedding, referred to as a c-vector. This combination uses a 2-dimensional (2D) self-attentive structure, which extends the standard self-attentive layer by averaging not only across time but also across different types of embeddings. Two types of 2D self-attentive structure in this paper are the simultaneous combination and the consecutive combination, adopting a single and multiple self-attentive layers respectively. The penalty term in the original self-attentive layer which is jointly minimised with the objective function to encourage diversity of annotation vectors is also modified to obtain not only different local peaks but also the overall trends in the multiple annotation vectors. Experiments on the AMI meeting corpus show that our modified penalty term improves the d- vector relative speaker error rate (SER) by 6% and 21% for d-vector systems, and a 10% further relative SER reduction can be obtained using the c-vector from our best 2D self-attentive structure.

up
0 users have voted:

Paper Details

Authors:
Guangzhi Sun, Chao Zhang, Phil Woodland
Submitted On:
12 May 2019 - 11:10am
Short Link:
Type:
Presentation Slides
Event:
Presenter's Name:
Guangzhi Sun
Paper Code:
SLP-L7.1
Document Year:
2019
Cite

Document Files

DiarisationPresentation3.pdf

(29)

Subscribe

[1] Guangzhi Sun, Chao Zhang, Phil Woodland, "Speaker Diarisation Using 2D Self-attentive Combination of Embeddings", IEEE SigPort, 2019. [Online]. Available: http://sigport.org/4465. Accessed: Nov. 15, 2019.
@article{4465-19,
url = {http://sigport.org/4465},
author = {Guangzhi Sun; Chao Zhang; Phil Woodland },
publisher = {IEEE SigPort},
title = {Speaker Diarisation Using 2D Self-attentive Combination of Embeddings},
year = {2019} }
TY - EJOUR
T1 - Speaker Diarisation Using 2D Self-attentive Combination of Embeddings
AU - Guangzhi Sun; Chao Zhang; Phil Woodland
PY - 2019
PB - IEEE SigPort
UR - http://sigport.org/4465
ER -
Guangzhi Sun, Chao Zhang, Phil Woodland. (2019). Speaker Diarisation Using 2D Self-attentive Combination of Embeddings. IEEE SigPort. http://sigport.org/4465
Guangzhi Sun, Chao Zhang, Phil Woodland, 2019. Speaker Diarisation Using 2D Self-attentive Combination of Embeddings. Available at: http://sigport.org/4465.
Guangzhi Sun, Chao Zhang, Phil Woodland. (2019). "Speaker Diarisation Using 2D Self-attentive Combination of Embeddings." Web.
1. Guangzhi Sun, Chao Zhang, Phil Woodland. Speaker Diarisation Using 2D Self-attentive Combination of Embeddings [Internet]. IEEE SigPort; 2019. Available from : http://sigport.org/4465