Hybrid-Module Transformer: enhancing speech emotion recognition with HuBERT, LSTM, and ResNet-50

Speech emotion recognition (SER) is a challenging task that involves identifying human emotions from speech. Traditional sequence models like recurrent neural network (RNN) and long short-term memory (LSTM) are limited by vanishing gradients and difficulty in capturing long-range dependencies. This...

Descrizione completa

Dettagli Bibliografici

Autori principali:	Xindong Huang, Wuhui Lin, Maming Chen, Hua Shi
Natura:	Artigo
Lingua:	Inglês
Pubblicazione:	PeerJ Inc. 2025-10-01
Serie:	PeerJ Computer Science
Soggetti:	Speech emotion recognition Deep learning HuBERT NLP
Accesso online:	https://peerj.com/articles/cs-3292.pdf
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!