Hybrid-Module Transformer: enhancing speech emotion recognition with HuBERT, LSTM, and ResNet-50
Bibliographic record
Abstract
Speech emotion recognition (SER) is a challenging task that involves identifying human emotions from speech. Traditional sequence models like recurrent neural network (RNN) and long short-term memory (LSTM) are limited by vanishing gradients and difficulty in capturing long-range dependencies. This article presents a novel model based on the Hybrid-Module-Transformer, which leverages the capabilities of Transformer modules to extract feature representations effectively, even with limited data. The model combines the strengths of Hidden-Unit BERT (HuBERT), LSTM, and Residual Network (ResNet-50) to achieve superior performance in speech emotion classification tasks. In the model, we utilized Mel-frequency cepstral coefficients (MFCC) and Spectrogram for feature extraction. Then, a HuBERT-LSTM framework is used to perform both speech-to-text recognition and emotion classification. We evaluate the model on two benchmark datasets: Ryerson Audio-Visual Database of Emotional Speech and Song (RAVDESS) and Multimodal EmotionLines Dataset (MELD). On the RAVDESS dataset, the model achieves a maximum accuracy of 76% and precision of 78%, while on the more challenging MELD dataset, it attains an accuracy of 72.9% and precision of 72.3%. These results demonstrate the effectiveness and generalizability of our model in both controlled and real-world conversational scenarios, making it a competitive solution for robust speech emotion recognition.
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 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 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; 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".