We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computation and Language

Title: Investigation of Densely Connected Convolutional Networks with Domain Adversarial Learning for Noise Robust Speech Recognition

Abstract: We investigate densely connected convolutional networks (DenseNets) and their extension with domain adversarial training for noise robust speech recognition. DenseNets are very deep, compact convolutional neural networks which have demonstrated incredible improvements over the state-of-the-art results in computer vision. Our experimental results reveal that DenseNets are more robust against noise than other neural network based models such as deep feed forward neural networks and convolutional neural networks. Moreover, domain adversarial learning can further improve the robustness of DenseNets against both, known and unknown noise conditions.
Comments: 7 pages, 5 figures, The 30th Conference on Electronic Speech Signal Processing (ESSV2019)
Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
Cite as: arXiv:2112.10108 [cs.CL]
  (or arXiv:2112.10108v1 [cs.CL] for this version)

Submission history

From: Chia-Yu Li [view email]
[v1] Sun, 19 Dec 2021 10:29:17 GMT (440kb,D)

Link back to: arXiv, form interface, contact.