Exploring the variable efficacy of Google speech-to-text with spontaneous bilingual speech in Cantonese and English
Bibliographic record
Abstract
With the growth of Automatic Speech Recognition (ASR) and voice user interface software, it is important to test for efficacy across different language varieties and identify sources of bias. Recent work assessing ASR efficacy and bias implicates factors like race, gender, dialect, and age as leading to different efficacy rates. Multilingualism presents another source of variation that ASR systems must grapple with, ranging from code-switching to phonetic variation both within and across speakers. Thus, variable ASR performance is likely exacerbated for multilingual communities. Using a spontaneous Cantonese-English bilingual speech corpus (Johnson, 2021), this study tests the efficacy of Google Speech-To-Text (STT) language models (Canadian English and Hong Kong Cantonese) with a heterogeneous bilingual speech community. Efficacy is assessed via fuzzy string matching between the STT and the manually corrected transcripts. STT performance is variable but overall better for English. Confidence ratings and matching scores are evaluated alongside listener ratings of perceived accentedness and various demographic groupings. The results of this study will provide practical guidance for using STT in the context of speech production research pipelines while highlighting its drawbacks concerning bias in a relatively understudied, multilingual group.
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 | 0.009 | 0.039 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.001 | 0.002 |
| 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".