Abstract

Owing to the suprasegmental behavior of emotional speech, turn-level features have demonstrated a better success than frame-level features for recognition-related tasks. Conventionally, such features are obtained via a brute-force collection of statistics over frames, thereby losing important local information in the process which affects the performance. To overcome these limitations, a novel feature extraction approach using latent topic models (LTMs) is presented in this study. Speech is assumed to comprise of a mixture of emotion-specific topics, where the latter capture emotionally salient information from the co-occurrences of frame-level acoustic features and yield better descriptors. Specifically, a supervised replicated softmax model (sRSM), based on restricted Boltzmann machines and distributed representations, is proposed to learn naturally discriminative topics. The proposed features are evaluated for the recognition of categorical or continuous emotional attributes via within and cross-corpus experiments conducted over acted and spontaneous expressions. In a within-corpus scenario, sRSM outperforms competing LTMs, while obtaining a significant improvement of 16.75% over popular statistics-based turn-level features for valence-based classification, which is considered to be a difficult task using only speech. Further analyses with respect to the turn duration show that the improvement is even more significant, 35%, on longer turns (>6 s), which is highly desirable for current turn-based practices. In a cross-corpus scenario, two novel adaptation-based approaches, instance selection, and weight regularization are proposed to reduce the inherent bias due to varying annotation procedures and cultural perceptions across databases. Experimental results indicate a natural, yet less severe, deterioration in performance - only 2.6% and 2.7%, thereby highlighting the generalization ability of the proposed features.

Original languageEnglish (US)
JournalEurasip Journal on Audio, Speech, and Music Processing
Volume2015
Issue number1
DOIs
StatePublished - 2015

Fingerprint

emotions
Speech recognition
Statistics
statistics
annotations
deterioration
pattern recognition
Deterioration
Feature extraction
Acoustics
occurrences
valence
acoustics
Experiments

Keywords

  • Cross-corpus
  • Speech emotion recognition
  • Suprasegmental features
  • Topic models

ASJC Scopus subject areas

  • Acoustics and Ultrasonics
  • Electrical and Electronic Engineering

Cite this

@article{0df4834f814c450999a30ed16090f038,
title = "Within and cross-corpus speech emotion recognition using latent topic model-based features",
abstract = "Owing to the suprasegmental behavior of emotional speech, turn-level features have demonstrated a better success than frame-level features for recognition-related tasks. Conventionally, such features are obtained via a brute-force collection of statistics over frames, thereby losing important local information in the process which affects the performance. To overcome these limitations, a novel feature extraction approach using latent topic models (LTMs) is presented in this study. Speech is assumed to comprise of a mixture of emotion-specific topics, where the latter capture emotionally salient information from the co-occurrences of frame-level acoustic features and yield better descriptors. Specifically, a supervised replicated softmax model (sRSM), based on restricted Boltzmann machines and distributed representations, is proposed to learn naturally discriminative topics. The proposed features are evaluated for the recognition of categorical or continuous emotional attributes via within and cross-corpus experiments conducted over acted and spontaneous expressions. In a within-corpus scenario, sRSM outperforms competing LTMs, while obtaining a significant improvement of 16.75{\%} over popular statistics-based turn-level features for valence-based classification, which is considered to be a difficult task using only speech. Further analyses with respect to the turn duration show that the improvement is even more significant, 35{\%}, on longer turns (>6 s), which is highly desirable for current turn-based practices. In a cross-corpus scenario, two novel adaptation-based approaches, instance selection, and weight regularization are proposed to reduce the inherent bias due to varying annotation procedures and cultural perceptions across databases. Experimental results indicate a natural, yet less severe, deterioration in performance - only 2.6{\%} and 2.7{\%}, thereby highlighting the generalization ability of the proposed features.",
keywords = "Cross-corpus, Speech emotion recognition, Suprasegmental features, Topic models",
author = "Mohit Shah and Chaitali Chakrabarti and Andreas Spanias",
year = "2015",
doi = "10.1186/s13636-014-0049-y",
language = "English (US)",
volume = "2015",
journal = "Eurasip Journal on Audio, Speech, and Music Processing",
issn = "1687-4714",
publisher = "Springer Publishing Company",
number = "1",

}

TY - JOUR

T1 - Within and cross-corpus speech emotion recognition using latent topic model-based features

AU - Shah, Mohit

AU - Chakrabarti, Chaitali

AU - Spanias, Andreas

PY - 2015

Y1 - 2015

N2 - Owing to the suprasegmental behavior of emotional speech, turn-level features have demonstrated a better success than frame-level features for recognition-related tasks. Conventionally, such features are obtained via a brute-force collection of statistics over frames, thereby losing important local information in the process which affects the performance. To overcome these limitations, a novel feature extraction approach using latent topic models (LTMs) is presented in this study. Speech is assumed to comprise of a mixture of emotion-specific topics, where the latter capture emotionally salient information from the co-occurrences of frame-level acoustic features and yield better descriptors. Specifically, a supervised replicated softmax model (sRSM), based on restricted Boltzmann machines and distributed representations, is proposed to learn naturally discriminative topics. The proposed features are evaluated for the recognition of categorical or continuous emotional attributes via within and cross-corpus experiments conducted over acted and spontaneous expressions. In a within-corpus scenario, sRSM outperforms competing LTMs, while obtaining a significant improvement of 16.75% over popular statistics-based turn-level features for valence-based classification, which is considered to be a difficult task using only speech. Further analyses with respect to the turn duration show that the improvement is even more significant, 35%, on longer turns (>6 s), which is highly desirable for current turn-based practices. In a cross-corpus scenario, two novel adaptation-based approaches, instance selection, and weight regularization are proposed to reduce the inherent bias due to varying annotation procedures and cultural perceptions across databases. Experimental results indicate a natural, yet less severe, deterioration in performance - only 2.6% and 2.7%, thereby highlighting the generalization ability of the proposed features.

AB - Owing to the suprasegmental behavior of emotional speech, turn-level features have demonstrated a better success than frame-level features for recognition-related tasks. Conventionally, such features are obtained via a brute-force collection of statistics over frames, thereby losing important local information in the process which affects the performance. To overcome these limitations, a novel feature extraction approach using latent topic models (LTMs) is presented in this study. Speech is assumed to comprise of a mixture of emotion-specific topics, where the latter capture emotionally salient information from the co-occurrences of frame-level acoustic features and yield better descriptors. Specifically, a supervised replicated softmax model (sRSM), based on restricted Boltzmann machines and distributed representations, is proposed to learn naturally discriminative topics. The proposed features are evaluated for the recognition of categorical or continuous emotional attributes via within and cross-corpus experiments conducted over acted and spontaneous expressions. In a within-corpus scenario, sRSM outperforms competing LTMs, while obtaining a significant improvement of 16.75% over popular statistics-based turn-level features for valence-based classification, which is considered to be a difficult task using only speech. Further analyses with respect to the turn duration show that the improvement is even more significant, 35%, on longer turns (>6 s), which is highly desirable for current turn-based practices. In a cross-corpus scenario, two novel adaptation-based approaches, instance selection, and weight regularization are proposed to reduce the inherent bias due to varying annotation procedures and cultural perceptions across databases. Experimental results indicate a natural, yet less severe, deterioration in performance - only 2.6% and 2.7%, thereby highlighting the generalization ability of the proposed features.

KW - Cross-corpus

KW - Speech emotion recognition

KW - Suprasegmental features

KW - Topic models

UR - http://www.scopus.com/inward/record.url?scp=84961290521&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84961290521&partnerID=8YFLogxK

U2 - 10.1186/s13636-014-0049-y

DO - 10.1186/s13636-014-0049-y

M3 - Article

AN - SCOPUS:84961290521

VL - 2015

JO - Eurasip Journal on Audio, Speech, and Music Processing

JF - Eurasip Journal on Audio, Speech, and Music Processing

SN - 1687-4714

IS - 1

ER -