Smart Manufacturing in Industrial AIoT 5.0 Applications: A Speech Emotion Recognition Approach

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F60076658%3A12310%2F25%3A43911222" target="_blank" >RIV/60076658:12310/25:43911222 - isvavai.cz</a>

Výsledek na webu

<a href="https://ieeexplore.ieee.org/abstract/document/11105780" target="_blank" >https://ieeexplore.ieee.org/abstract/document/11105780</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1109/JIOT.2025.3594566" target="_blank" >10.1109/JIOT.2025.3594566</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Smart Manufacturing in Industrial AIoT 5.0 Applications: A Speech Emotion Recognition Approach

Popis výsledku v původním jazyce

In the era of Industrial Internet of Things (IIoT) 5.0, recognizing emotions through speech plays a crucial role in creating advanced and emotionally intelligent systems for better human-machine interactions (HMIs) for various IoT applications. These systems are especially valuable in smart manufacturing environments and their respective applications. The proposed research showcase the challenge of improving speech quality for reliable speech emotion recognition (SER) in noisy industrial settings by introducing an improved modulation spectral subtraction (IMSS) method. The IMSS technique enhances traditional analysis-modification-synthesis (AMS) frameworks with refined processing in the modulation domain, utilizing advanced noise estimation approaches like the minimum statistics (MS) method. To recognize emotions, the study employs a machine learning algorithm based on a convolutional neural network (CNN). The proposed algorithm processes the enhanced speech signals to accurately detect emotional states in speech. The combination of the IMSS method with the CNN model ensures that emotional details in speech are retained, which is essential for effective SER. The proposed technique significantly improves speech clarity and quality, evaluated through objective measures, such as the perceptual evaluation of speech quality (PESQ). The experimental results show notable improvements in speech quality, with an average 14.91% increase in PESQ scores for input signal-to-noise ratios (SNRs) between 0 and 15 dB, along with a 63% reduction in Log Spectral Distance. These findings highlight the method&apos;s effectiveness under various noisy conditions. Spectrogram analysis further demonstrates the IMSS method&apos;s ability to enhance the accuracy and reliability of SER, which is reinforced by the strong performance of the CNN in classification tasks. By optimizing the modulation frame duration to 128 ms, the work approach provides a valuable contribution to adaptive and safety-oriented IIoT 5.0 applications.

Název v anglickém jazyce

Smart Manufacturing in Industrial AIoT 5.0 Applications: A Speech Emotion Recognition Approach

Popis výsledku anglicky

In the era of Industrial Internet of Things (IIoT) 5.0, recognizing emotions through speech plays a crucial role in creating advanced and emotionally intelligent systems for better human-machine interactions (HMIs) for various IoT applications. These systems are especially valuable in smart manufacturing environments and their respective applications. The proposed research showcase the challenge of improving speech quality for reliable speech emotion recognition (SER) in noisy industrial settings by introducing an improved modulation spectral subtraction (IMSS) method. The IMSS technique enhances traditional analysis-modification-synthesis (AMS) frameworks with refined processing in the modulation domain, utilizing advanced noise estimation approaches like the minimum statistics (MS) method. To recognize emotions, the study employs a machine learning algorithm based on a convolutional neural network (CNN). The proposed algorithm processes the enhanced speech signals to accurately detect emotional states in speech. The combination of the IMSS method with the CNN model ensures that emotional details in speech are retained, which is essential for effective SER. The proposed technique significantly improves speech clarity and quality, evaluated through objective measures, such as the perceptual evaluation of speech quality (PESQ). The experimental results show notable improvements in speech quality, with an average 14.91% increase in PESQ scores for input signal-to-noise ratios (SNRs) between 0 and 15 dB, along with a 63% reduction in Log Spectral Distance. These findings highlight the method&apos;s effectiveness under various noisy conditions. Spectrogram analysis further demonstrates the IMSS method&apos;s ability to enhance the accuracy and reliability of SER, which is reinforced by the strong performance of the CNN in classification tasks. By optimizing the modulation frame duration to 128 ms, the work approach provides a valuable contribution to adaptive and safety-oriented IIoT 5.0 applications.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

—

Návaznosti

I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2025

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

IEEE INTERNET OF THINGS JOURNAL

ISSN

2327-4662

e-ISSN

—

Svazek periodika

12

Číslo periodika v rámci svazku

20

Stát vydavatele periodika

US - Spojené státy americké

Počet stran výsledku

9

Strana od-do

42693-42701

Kód UT WoS článku

001589934500002

EID výsledku v databázi Scopus

—