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

Download:

Current browse context:

cs.CV

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

Electrical Engineering and Systems Science > Audio and Speech Processing

Title: Speaker embeddings by modeling channel-wise correlations

Abstract: Speaker embeddings extracted with deep 2D convolutional neural networks are typically modeled as projections of first and second order statistics of channel-frequency pairs onto a linear layer, using either average or attentive pooling along the time axis. In this paper we examine an alternative pooling method, where pairwise correlations between channels for given frequencies are used as statistics. The method is inspired by style-transfer methods in computer vision, where the style of an image, modeled by the matrix of channel-wise correlations, is transferred to another image, in order to produce a new image having the style of the first and the content of the second. By drawing analogies between image style and speaker characteristics, and between image content and phonetic sequence, we explore the use of such channel-wise correlations features to train a ResNet architecture in an end-to-end fashion. Our experiments on VoxCeleb demonstrate the effectiveness of the proposed pooling method in speaker recognition.
Comments: Accepted at Interspeech 2021
Subjects: Audio and Speech Processing (eess.AS); Computer Vision and Pattern Recognition (cs.CV)
Cite as: arXiv:2104.02571 [eess.AS]
  (or arXiv:2104.02571v2 [eess.AS] for this version)

Submission history

From: Themos Stafylakis [view email]
[v1] Tue, 6 Apr 2021 15:10:14 GMT (265kb)
[v2] Wed, 7 Jul 2021 14:00:22 GMT (265kb)

Link back to: arXiv, form interface, contact.