Next Article in Journal
Research on Centroid Distribution and Dynamic Characteristics of Irregular Tooth End Milling Cutters
Previous Article in Journal
Spatial Analysis for Landscape Changes: A Bibliometric Review
Article

U-Vectors: Generating Clusterable Speaker Embedding from Unlabeled Data

1
Department of Computer Science & Engineering, Bangladesh University of Business & Technology, Dhaka 1216, Bangladesh
2
Department of Information Technology, Faculty of Computing & Information Technology, King Abdulaziz University, Jeddah 21589, Saudi Arabia
3
Department of Computer Science and Engineering, University of Asia Pacific, Dhaka 1216, Bangladesh
4
Department of Computer Science and Engineering, University of Aizu, Aizu-Wakamatsu 965-8580, Japan
*
Author to whom correspondence should be addressed.
Academic Editor: Byung-Gyu Kim
Appl. Sci. 2021, 11(21), 10079; https://doi.org/10.3390/app112110079
Received: 1 October 2021 / Revised: 22 October 2021 / Accepted: 22 October 2021 / Published: 27 October 2021
(This article belongs to the Section Computing and Artificial Intelligence)
Speaker recognition deals with recognizing speakers by their speech. Most speaker recognition systems are built upon two stages, the first stage extracts low dimensional correlation embeddings from speech, and the second performs the classification task. The robustness of a speaker recognition system mainly depends on the extraction process of speech embeddings, which are primarily pre-trained on a large-scale dataset. As the embedding systems are pre-trained, the performance of speaker recognition models greatly depends on domain adaptation policy, which may reduce if trained using inadequate data. This paper introduces a speaker recognition strategy dealing with unlabeled data, which generates clusterable embedding vectors from small fixed-size speech frames. The unsupervised training strategy involves an assumption that a small speech segment should include a single speaker. Depending on such a belief, a pairwise constraint is constructed with noise augmentation policies, used to train AutoEmbedder architecture that generates speaker embeddings. Without relying on domain adaption policy, the process unsupervisely produces clusterable speaker embeddings, termed unsupervised vectors (u-vectors). The evaluation is concluded in two popular speaker recognition datasets for English language, TIMIT, and LibriSpeech. Also, a Bengali dataset is included to illustrate the diversity of the domain shifts for speaker recognition systems. Finally, we conclude that the proposed approach achieves satisfactory performance using pairwise architectures. View Full-Text
Keywords: speaker recognition; clustering; twin networks; deep learning speaker recognition; clustering; twin networks; deep learning
Show Figures

Figure 1

MDPI and ACS Style

Mridha, M.F.; Ohi, A.Q.; Monowar, M.M.; Hamid, M.A.; Islam, M.R.; Watanobe, Y. U-Vectors: Generating Clusterable Speaker Embedding from Unlabeled Data. Appl. Sci. 2021, 11, 10079. https://doi.org/10.3390/app112110079

AMA Style

Mridha MF, Ohi AQ, Monowar MM, Hamid MA, Islam MR, Watanobe Y. U-Vectors: Generating Clusterable Speaker Embedding from Unlabeled Data. Applied Sciences. 2021; 11(21):10079. https://doi.org/10.3390/app112110079

Chicago/Turabian Style

Mridha, Muhammad F., Abu Q. Ohi, Muhammad M. Monowar, Md. A. Hamid, Md. R. Islam, and Yutaka Watanobe. 2021. "U-Vectors: Generating Clusterable Speaker Embedding from Unlabeled Data" Applied Sciences 11, no. 21: 10079. https://doi.org/10.3390/app112110079

Find Other Styles
Note that from the first issue of 2016, MDPI journals use article numbers instead of page numbers. See further details here.

Article Access Map by Country/Region

1
Back to TopTop