Deep Learning-Based Speech Emotion Recognition for IoT Edge Devices: A Comparative Study
Bibliographic record
Abstract
With advancements in artificial intelligence (AI), particularly in pattern recognition, significant progress has been made in recognising human emotions from speech characteristics, facial activity, and physiological responses. However, the expansion of Internet of Things (IoT)-based infrastructures has increased pressure on conventional cloud systems due to the high volume of transmitted data and the need for real-time responsiveness. As a remedy, edge computing has emerged as a distributed alternative, enabling localised data processing and reducing dependency on remote servers. In this context, the present study evaluates the classification performance of three hybrid deep learning (DL) models—Convolutional Neural Network–Dense Neural Network (CNN-Dense), Long Short-Term Memory–Convolutional Neural Network (LSTM-CNN), and Dense–Long Short-Term Memory (Dense-LSTM) —within a simulated edge-based environment. The Toronto Emotional Speech Set (TESS) dataset was employed, and experimental workflows were implemented via Amazon Web Services (AWS) to simulate edge resource limitations. Accuracy was assessed using macro-averaged metrics, including precision, recall, and F1-score. Among the models, CNN-Dense showed the highest performance, achieving an F1-score of 96%, followed by LSTM-CNN (95%) and Dense-LSTM (93%). The findings suggest that CNN–Dense may offer feature extraction advantages, and that hybrid models could be promising for emotion classification in decentralised systems.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.001 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".