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


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


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

Computer Science > Sound

Title: Variational Autoencoders for Learning Latent Representations of Speech Emotion: A Preliminary Study

Abstract: Learning the latent representation of data in unsupervised fashion is a very interesting process that provides relevant features for enhancing the performance of a classifier. For speech emotion recognition tasks, generating effective features is crucial. Currently, handcrafted features are mostly used for speech emotion recognition, however, features learned automatically using deep learning have shown strong success in many problems, especially in image processing. In particular, deep generative models such as Variational Autoencoders (VAEs) have gained enormous success for generating features for natural images. Inspired by this, we propose VAEs for deriving the latent representation of speech signals and use this representation to classify emotions. To the best of our knowledge, we are the first to propose VAEs for speech emotion classification. Evaluations on the IEMOCAP dataset demonstrate that features learned by VAEs can produce state-of-the-art results for speech emotion classification.
Comments: Proc. Interspeech 2018
Subjects: Sound (cs.SD); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS); Machine Learning (stat.ML)
Cite as: arXiv:1712.08708 [cs.SD]
  (or arXiv:1712.08708v3 [cs.SD] for this version)

Submission history

From: Siddique Latif [view email]
[v1] Sat, 23 Dec 2017 03:54:00 GMT (38kb)
[v2] Mon, 26 Mar 2018 07:47:54 GMT (373kb,D)
[v3] Tue, 28 Jul 2020 01:35:27 GMT (373kb,D)

Link back to: arXiv, form interface, contact.