Development of an advanced ai-based model for human psychoemotional state analysis

Eastern-European Journal of Enterprise Technologies, Journal Year: 2023, Volume and Issue: 6(4 (126)), P. 39 - 49

Published: Dec. 28, 2023

The research focuses on developing a novel method for the automatic recognition of human psychoemotional states (PES) using deep learning technology. This is centered analyzing speech signals to classify distinct emotional states. primary challenge addressed by this accurately perform multiclass classification seven states, namely joy, fear, anger, sadness, disgust, surprise, and neutral state. Traditional methods have struggled distinguish these complex nuances in speech. study successfully developed model capable extracting informative features from audio recordings, specifically mel spectrograms mel-frequency cepstral coefficients. These were then used train two convolutional neural networks, resulting classifier model. uniqueness lies its use dual-feature approach employment networks classification. has demonstrated high accuracy emotion recognition, with an rate 0.93 validation subset. effectiveness can be attributed comprehensive synergistic coefficients, which provide more nuanced analysis expressions presented broad applicability various domains, including enhancing human-machine interface interactions, implementation aviation industry, healthcare, marketing, other fields where understanding emotions through crucial

Language: Английский

---

Comparative Analysis of Advanced ML Techniques for Urdu Speech Emotion Recognition: An Acoustic Study in Low-Resource Languages*

2021 7th Annual International Conference on Network and Information Systems for Computers (ICNISC), Journal Year: 2023, Volume and Issue: unknown

Published: Oct. 27, 2023

Language: Английский

---

Capturing Time Dynamics From Speech Using Neural Networks for Surgical Mask Detection

IEEE Journal of Biomedical and Health Informatics, Journal Year: 2022, Volume and Issue: 26(8), P. 4291 - 4302

Published: May 6, 2022

The importance of detecting whether a person wears face mask while speaking has tremendously increased since the outbreak SARS-CoV-2 (COVID-19), as wearing can help to reduce spread virus and mitigate public health crisis. Besides affecting human speech characteristics related frequency, masks cause temporal interferences in speech, altering pace, rhythm, pronunciation speed. In this regard, paper presents two effective neural network models detect surgical from audio. proposed architectures are both based on Convolutional Neural Networks (CNNs), chosen an optimal approach for spatial processing audio signals. One architecture applies Long Short-Term Memory (LSTM) model time-dependencies. Through additional attention mechanism, LSTM-based enables extraction more salient information. other (named ConvTx) retrieves relative position sequence through positional encoder transformer module. order assess which extent complement each when modelling dynamics, we also explore combination LSTM Transformers three hybrid models. Finally, investigate data augmentation techniques, such as, using transitions between frames considering gender-dependent frameworks might impact performance architectures. Our experimental results show that one achieves best performance, surpassing existing state-of-the-art task at hand.

Language: Английский

---

Automatic Emotion Recognition System: A cross Culture Study between Tamil and Russian Speaking Children

Published: Feb. 23, 2023

Speech based communication contains both linguistic and paralinguistic information. It conveys emotions not only through its verbal communication, but also voice characteristics such as pitch, volume, stress. The objective of this study is to convey the child's emotion a caretaker. To achieve this, cross-cultural four "joy - neutral (calm state) sadness anger" in children aged 8-12 years considered. data collection carried out multiple modes like acting speech Russian spontaneous Tamil language. SVM MLP classitiers were used by specialist train validate languages. accuracy achieved team 84.1% Indian specialists CNN obtained training validation 90% >68% over 75 Epoch 7-layer 95% > 65% 50 Epoch. goal see how well can recognise emotional condition on their speech, results are correlated with computer-based classification algorithms various machine learning deep net architectures.

Language: Английский

---

The Problematics of Gender for Aviation Emergency Communication during an Inflight Emergency: A Case Study

Qualitative Sociology Review, Journal Year: 2023, Volume and Issue: 19(2), P. 6 - 29

Published: April 30, 2023

Due to the rarity of female pilots, aviation communication is typically conducted in a single-gender environment. The role gender interactions during inflight emergencies has not yet been adequately explored. This single case analysis uses qualitative approach based on conversation analytic transcripts investigate how may be relevant either explicitly or implicitly radio transmissions between flight crew and Air Traffic Control (ATC) personnel, as well internal ATC phone participants work handle an emergency. incident involved pilot male copilot, thus providing naturally occurring rare event explore potential relevance gender. shows that explicit references are limited occasional asymmetrical use gendered address terms pronouns. Participants also used interactional formulations that—while gendered—have associated previous research with differences interaction, for example, indirect forms requests complaints, actions imply inferences about emotional state participants, possible confusion over identity given transitions sounding voices speaking behalf plane. findings discussed implications can impact emergency incidents.

Language: Английский

---