The RACAD speech corpus of New Brunswick Acadian French: Design and applications
Bibliographic record
Abstract
The RACAD (Reconnaissance automatique de l 'acadien) speech corpus contains high quality audio recordings that can be used to develop recognition systems for the regional varieties of French spoken in the province of New Brunswick, Canada.Its design is informed by linguistic analyses of Acadian French.The corpus contains sentences read by 140 speakers who were selected according to age, gender and geographical region.This paper presents a preliminary application of the corpus in automatic speech recognition research; it outlines an original global monophone recognition model that is designed to handle linguistic variability.Global phone and word recognition rates for this model are satisfactory (about 90%), but they vary considerably across geographical locations.Possible applications of the RACAD corpus in acoustic phonetic and socio-phonetic studies of dialect variation are also described in this paper. R É SU M ÉDans le but de développer des systèmes de reconnaissance automatique des variétés de français parlées dans la province du Nouveau-Brunswick, au Canada, un corpus d 'enregistrements de haute qualité, le corpus RACAD (Reconnaissance automatique de l 'acadien), a été recueilli.Ce corpus est constitué de phrases lues par 140 locuteurs.Suivant la méthodologie employée dans les études linguistiques portant sur le français acadien, les locuteurs ont été sélectionnés d 'après leur âge, leur sexe et leur appartenance géographique.Cet article décrit une première application du processus de reconnaissance automatique de la parole à partir de ce corpus; il présente un modèle monophone global qui tient compte de la variabilité linguistique dans le RACAD.Les résultats montrent que les taux de reconnaissance globale des phones et des mots sont satisfaisants (environ 90%), mais que ces taux varient entre les diverses régions géographiques.Des applications possibles du RACAD, dans des analyses de phonétique acoustique et de sociophonétique de la variation régionale, sont aussi décrites dans le présent article.Sound & V ib ra tio n In s tru m e n ta tio n a n d Engineering VAVW.scantekinc.com
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 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.000 |
| Insufficient payload (model declined to judge) | 0.011 | 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".