Johnson O. S.

A Reproducible Hybrid LSTM–SVM Framework for Speech Emotion Recognition Using Temporal–Acoustic Embeddings

Speech emotion recognition (SER) seeks to infer affective states from paralinguistic cues embedded in voice, but reliable evaluation is difficult because emotional speech is sequential, speaker-dependent, and sensitive to corpus and validation design. This study presents a hybrid Long Short-Term Memory–Support Vector M...