RBi-LSTM Based CNN Model for Speech Emotion Recognition
Bibliographic record
Abstract
Speech Emotion Recognition (SER) has drawn considerable interest from the scientific community because of its promising applications in human-computer interaction and affective computing. The work in this paper is aimed at designing and testing a high-performance deep learning architecture for emotion recognition from speech signals. Using a combined dataset of the popular Ryerson Audio-Visual Database of Emotional Speech and Song (RAVDESS) and the Toronto Emotional Speech Set (TESS), we used a holistic feature extraction method, extracting prominent acoustic features using Mel-Frequency Cepstral Coefficients (MFCCs) with derivatives, Chroma features, Mel Spectrograms, Zero Crossing Rate (ZCR), and Root Mean Square (RMS) energy. The system's backbone is a hybrid neural network design combining Convolutional Neural Networks (CNNs) to extract local spectro-temporal features, and stacked Bidirectional Long Short-Term Memory networks with residual connections (ResBiLSTMs) for effective modeling of long-term temporal relationships. Deployed and trained in the Google Colab environment, the model proved highly effective when tested on a held-out test set from the merged dataset and performed a high accuracy of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 6. 8 3 \%}$</tex>. This finding confirms the efficiency of the selected feature set and the particular CNN-Residual BiLSTM architecture for robust SER, especially when taking advantage of the additional data diversity of merging typical datasets. In contrast to previous CNN-BiLSTM methods, our model combines a variety of acoustic features into a single sequence representation and incorporates residual skip connections to enhance gradient flow. In comparison to models trained on a single corpus, we further evaluate on a combined RAVDESS+TESS dataset, showing better generalisation.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.001 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.001 | 0.000 |
| Insufficient payload (model declined to judge) | 0.009 | 0.002 |
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; both teacher heads agree on what is shown here.
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".