Autors: Pleshkova, S. G., Bekiarski, A. B.
Title: Speaker Identification Algorithm Using Biometric Speech Features in Deep Learning Neural Network
Keywords: Siamese deep learning neural network, speaker identification, speech features, voice biometric characteristics

Abstract: Speaker identification is based on the specific voice biometric characteristics comparison of two speaking persons to decide whether or not the voice of one called 'unknown' is similar to the other called 'known'. It is proposed in this article to develop speaker identification algorithm using Siamese deep learning neural network to compare the voice biometric characteristics of two speakers. The results achieved from algorithm implementation are presented to demonstrate the algorithm ability to speaker identification with sufficient precision.

References

  1. Y. Mohamed, N. Mohamed, and A. Anter, "Person identification is a very important method applied in forensic," In book: Nature-Inspired Method for Smart Healthcare and Medical Data, December, 2023, pp.1-24.
  2. Ch.Wan, P. Patimavirujh, Kr. Singhavorawuti, and D. Saelim, "Voice biometric," IT4363 Information system security and Auditting, January, 2016.
  3. M. Garg1, A. Arora, and S.Gupta, "A novel feature biometric fusion approach for iris, speech and signature", Computer methods in materials science, Vol. 20, 2020, No. 2.
  4. Y. Li, and L. Ph. Chen,"A Survey on Siamese Network: Methodologies, Applications and Opportunities ", IEEE Transactions on Artificial Intelligence, December 2022.
  5. Pr. Duraira, and S. Sriuppili, " Speech processing: MFCC feature extraction techniques - An investigation", Journal of Physics conference series 1717(1), January 2021.
  6. Open Speech and Languages Resources, https://openslr.org/
  7. TensorFlow.1D Convolutional network definition. https://www.tensorflow.org/api-docs/python/tf/keras/layers/Conv1D

Issue

2024 33rd International Scientific Conference Electronics, ET 2024 - Proceedings, 2024, , https://doi.org/10.1109/ET63133.2024.10721532

Вид: публикация в международен форум, публикация в реферирано издание, индексирана в Scopus