BERSting at the Screams: Recognition of Shouted and Distressed Speech from Smartphone Recordings
Bibliographic record
Abstract
People shout for different reasons: to greet from far away, to express extreme joy or anger. But when does shouting mean someone is in distress and needs help? How does distressed shouting differ from distressed talking? There has been extensive research into speech emotion recognition, however, these models are often trained on audio datasets that have optimally positioned microphones. We collected a novel dataset containing prosodically emotional speech from 96 professional actors. The data was recorded on their own smartphones with a variety of obstructions and at varying distances in their home. We have validated the data for a proposed use case to train a paralinguistic distressed speech detection model. To test the difficulty of the task, we used support vector machines to classify the data into distressed, not distressed, shouted, not shouted and a combination of these classes on the currently validated data. We used both a set of extracted acoustic features as a basic and interpretable representation, along with embeddings from wav2vec 2.0 (w2v2) as a state of the art representation as preliminary benchmarks for the dataset. The best results were achieved with the w2v2 representation, with an F1 score of 0.91 for the classification of shouting, 0.79 for distress and 0.70 for distressed shouting, suggesting a need for further research in this area.
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 | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.001 | 0.000 |
| 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".