eduzhai > Applied Sciences > Engineering >

Mask Detection and Breath Monitoring from Speech on Data Augmentation Feature Representation and Modeling

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 5 pages

Abstract: This paper introduces our approaches for the Mask and Breathing Sub-Challengein the Interspeech COMPARE Challenge 2020. For the mask detection task, wetrain deep convolutional neural networks with filter-bank energies,gender-aware features, and speaker-aware features. Support Vector Machinesfollows as the back-end classifiers for binary prediction on the extracted deepembeddings. Several data augmentation schemes are used to increase the quantityof training data and improve our models robustness, including speedperturbation, SpecAugment, and random erasing. For the speech breath monitoringtask, we investigate different bottleneck features based on the Bi-LSTMstructure. Experimental results show that our proposed methods outperform thebaselines and achieve 0.746 PCC and 78.8 UAR on the Breathing and Maskevaluation set, respectively.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×