Speech emotion recognition with light weight deep neural ensemble model using hand crafted features

Abstract Automatic emotion detection has become crucial in various domains, such as healthcare, neuroscience, smart home technologies, and human-computer interaction (HCI). Speech Emotion Recognition (SER) has attracted considerable attention because of its potential to improve conversational roboti...

Full description

Saved in:
Bibliographic Details
Main Authors: Jaher Hassan Chowdhury, Sheela Ramanna, Ketan Kotecha
Format: Article
Language:English
Published: Nature Portfolio 2025-04-01
Series:Scientific Reports
Subjects:
Online Access:https://doi.org/10.1038/s41598-025-95734-z
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849737785540345856
author Jaher Hassan Chowdhury
Sheela Ramanna
Ketan Kotecha
author_facet Jaher Hassan Chowdhury
Sheela Ramanna
Ketan Kotecha
author_sort Jaher Hassan Chowdhury
collection DOAJ
description Abstract Automatic emotion detection has become crucial in various domains, such as healthcare, neuroscience, smart home technologies, and human-computer interaction (HCI). Speech Emotion Recognition (SER) has attracted considerable attention because of its potential to improve conversational robotics and human-computer interaction (HCI) systems. Despite its promise, SER research faces challenges such as data scarcity, the subjective nature of emotions, and complex feature extraction methods. In this paper, we seek to investigate whether a lightweight deep neural ensemble model (CNN and CNN_Bi-LSTM) using well-known hand-crafted features such as ZCR, RMSE, Chroma STFT, and MFCC would outperform models that use automatic feature extraction techniques (e.g., spectrogram-based methods) on benchmarked datasets. The focus of this paper is on the effectiveness of careful fine-tuning of the neural models with learning rate (LR) schedulers and applying regularization techniques. Our proposed ensemble model is validated using five publicly available datasets: RAVDESS, TESS, SAVEE, CREMA-D, and EmoDB. Accuracy, AUC-ROC, AUC-PRC, and F1-score metrics were used for performance testing, and the LIME (Local Interpretable Model-agnostic Explanations) technique was used for interpreting the results of our proposed ensemble model. Results indicate that our ensemble model consistently outperforms individual models, as well as several compared models which include spectrogram-based models for the above datasets in terms of the evaluation metrics.
format Article
id doaj-art-efecc0b38d654728b60a1619d48a7397
institution DOAJ
issn 2045-2322
language English
publishDate 2025-04-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-efecc0b38d654728b60a1619d48a73972025-08-20T03:06:49ZengNature PortfolioScientific Reports2045-23222025-04-0115111410.1038/s41598-025-95734-zSpeech emotion recognition with light weight deep neural ensemble model using hand crafted featuresJaher Hassan Chowdhury0Sheela Ramanna1Ketan Kotecha2The University of WinnipegThe University of WinnipegSymbiosis International (Deemed University)Abstract Automatic emotion detection has become crucial in various domains, such as healthcare, neuroscience, smart home technologies, and human-computer interaction (HCI). Speech Emotion Recognition (SER) has attracted considerable attention because of its potential to improve conversational robotics and human-computer interaction (HCI) systems. Despite its promise, SER research faces challenges such as data scarcity, the subjective nature of emotions, and complex feature extraction methods. In this paper, we seek to investigate whether a lightweight deep neural ensemble model (CNN and CNN_Bi-LSTM) using well-known hand-crafted features such as ZCR, RMSE, Chroma STFT, and MFCC would outperform models that use automatic feature extraction techniques (e.g., spectrogram-based methods) on benchmarked datasets. The focus of this paper is on the effectiveness of careful fine-tuning of the neural models with learning rate (LR) schedulers and applying regularization techniques. Our proposed ensemble model is validated using five publicly available datasets: RAVDESS, TESS, SAVEE, CREMA-D, and EmoDB. Accuracy, AUC-ROC, AUC-PRC, and F1-score metrics were used for performance testing, and the LIME (Local Interpretable Model-agnostic Explanations) technique was used for interpreting the results of our proposed ensemble model. Results indicate that our ensemble model consistently outperforms individual models, as well as several compared models which include spectrogram-based models for the above datasets in terms of the evaluation metrics.https://doi.org/10.1038/s41598-025-95734-zSpeech emotion recognitionAveraging ensembleConvolutional neural networkBi-directional LSTMAudio signal processing
spellingShingle Jaher Hassan Chowdhury
Sheela Ramanna
Ketan Kotecha
Speech emotion recognition with light weight deep neural ensemble model using hand crafted features
Scientific Reports
Speech emotion recognition
Averaging ensemble
Convolutional neural network
Bi-directional LSTM
Audio signal processing
title Speech emotion recognition with light weight deep neural ensemble model using hand crafted features
title_full Speech emotion recognition with light weight deep neural ensemble model using hand crafted features
title_fullStr Speech emotion recognition with light weight deep neural ensemble model using hand crafted features
title_full_unstemmed Speech emotion recognition with light weight deep neural ensemble model using hand crafted features
title_short Speech emotion recognition with light weight deep neural ensemble model using hand crafted features
title_sort speech emotion recognition with light weight deep neural ensemble model using hand crafted features
topic Speech emotion recognition
Averaging ensemble
Convolutional neural network
Bi-directional LSTM
Audio signal processing
url https://doi.org/10.1038/s41598-025-95734-z
work_keys_str_mv AT jaherhassanchowdhury speechemotionrecognitionwithlightweightdeepneuralensemblemodelusinghandcraftedfeatures
AT sheelaramanna speechemotionrecognitionwithlightweightdeepneuralensemblemodelusinghandcraftedfeatures
AT ketankotecha speechemotionrecognitionwithlightweightdeepneuralensemblemodelusinghandcraftedfeatures