Strategies to Enhance Whispered Speech Speaker Verification: A Comparative Analysis
Bibliographic record
Abstract
Today, automated speech-enabled tools are increasingly being used in everyday environments. This mobility has created new challenges for developers, who are now faced with input speech of varying styles (e.g. whispered) and corrupted by different noise sources. In this paper, special emphasis is placed on whispered speech, an underexplored yet burgeoning area due to the rapid proliferation of smartphones around the world. More specifically, this paper explores the performance boundaries achievable with whispered speech for a speaker verification task, both in matched and mismatched train/test conditions. Several strategies are investigated to improve the performance in the mismatched scenario, as well as in situations involving ambient noise. Our results agree with previously reported studies in adjacent areas, that significant gains could be obtained by training speaker models with both naturally voiced and whispered speech data. Moreover, additional gains could be achieved with speaking style and gender dependent systems. Overall, speaker verification performance inline with that obtained with naturally-voiced speech could be attained for whispered speech once specific strategies were put in place. Particularly, feature fusion showed to be an important strategy for practical applications in both clean and noisy conditions.
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 | 0.004 | 0.007 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 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 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 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".