Speech Intelligibility Assessment using Automatic Speech Recognition
Bibliographic record
Abstract
Hearing loss affects approximately 1.59 billion individuals globally, with projections indicating that nearly 2.5 billion will be impacted by 2050. Despite the increasing prevalence, many individuals delay seeking help, even in well-resourced settings, leading to a significant gap between clinical diagnoses and self-reported difficulties. In Canada, while 19.4% of the population exhibits measurable hearing loss, only 3.7% perceive their impairment, highlighting the need for improved hearing assessment methodologies. A key challenge in hearing impairment is difficulty understanding speech in noise, which traditional pure-tone audiometry fails to capture effectively. This thesis investigates the integration of Automatic Speech Recognition (ASR) models into speech-in-noise (SiN) testing to enhance hearing aid evaluations and enable automated, scalable, and clinically relevant assessments. The research examines ASR models under diverse acoustic conditions, demonstrating their effectiveness in quantifying Signal-to-Noise Ratio (SNR) loss—an essential measure of functional hearing ability. Results show that ASR-based scoring aligns closely with audiologist evaluations, reinforcing the potential of these models to support clinical decisionmaking and improve access to reliable hearing assessments. A key outcome of this research is the development of an automated, two-version desktop graphical user interface (GUI) for administering SiN tests. This tool facilitates test playback, response recording, and real-time SNR loss computation while enabling seamless de-identified data uploads to cloud-based ASR services, such as Amazon Web Services (AWS) and Microsoft Azure. The study also explores the electroacoustic evaluation of hearing aids under various speech and noise configurations, including the impact of face masks, directional microphone settings, and reverberation levels. Findings reveal that ASR models can effectively process hearing aid test recordings without requiring clean reference signals, offering a more scalable alternative to traditional electroacoustic assessments. To further bridge accessibility gaps, a cross-platform mobile application was developed, integrating an on-device ASR model for self-administered SiN testing. The app enables individuals to assess their speech-in-noise performance remotely, supporting offline functionality for users in areas with limited internet access. Pilot testing with normal-hearing adults demonstrated that the mobile app reliably captures and processes SiN responses across different microphone configurations and loudspeaker setups, achieving performance comparable to cloud-based ASR solutions. This work contributes to the field of audiology by advancing ASR-driven hearing assessments, improving accessibility, and reducing reliance on clinic-based evaluations. By integrating ASR technologies into automated testing frameworks and mobile applications, this research lays the groundwork for more inclusive, efficient, and scalable solutions in hearing healthcare. These findings have direct implications for early intervention strategies, hearing aid optimization, and the broader adoption of tele-audiology solutions in real-world environments.
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 | 0.004 |
| 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 | 0.002 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.003 |
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".