HAL will be down for maintenance from Friday, June 10 at 4pm through Monday, June 13 at 9am. More information
Skip to Main content Skip to Navigation
Master Thesis

Modélisation auto-supervisée de la parole affective spontanée

Abstract : Self-supervised pre-trained models using unlabeled data to extract representations have been widely explored in the field of automatic speech processing. This paper explores a new approach of extracting linguistic representations from transcriptions using pre-trained self-supervised models for time-continuous emotion recognition. We examined an alignment method to fit linguistic representations with time. Experimental results show that linguistic self-supervised representations can predict emotions in arousal and valence dimensions and achieve excellent results as acoustic self-supervised representations.
Document type :
Master Thesis
Complete list of metadata

https://dumas.ccsd.cnrs.fr/dumas-03516512
Contributor : Uga - Llasic Sciences Du Langage & Fles Connect in order to contact the contributor
Submitted on : Friday, January 7, 2022 - 11:22:44 AM
Last modification on : Friday, January 21, 2022 - 4:14:01 AM
Long-term archiving on: : Friday, April 8, 2022 - 6:40:40 PM

File

TONG Ziyi_M2_IDL.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : dumas-03516512, version 1

Citation

Ziyi Tong. Modélisation auto-supervisée de la parole affective spontanée. Sciences de l'Homme et Société. 2022. ⟨dumas-03516512⟩

Share

Metrics

Record views

29

Files downloads

12