MétaCan
Menu
Back to cohort
Record W2984770459 · doi:10.1121/1.5137272

Improved vowel labeling for prenasal merger using customized forced alignment

2019· article· en· W2984770459 on OpenAlexaboutno aff
Yuanming Shi, Margaret E. L. Renwick, Frederick Maier

Bibliographic record

VenueThe Journal of the Acoustical Society of America · 2019
Typearticle
Languageen
FieldComputer Science
TopicSpeech Recognition and Synthesis
Canadian institutionsnot available
Fundersnot available
KeywordsPronunciationVowelFormantComputer scienceSpeech recognitionCentroidLinguisticsArtificial intelligence

Abstract

fetched live from OpenAlex

Forced alignment is a popular technique for gaining phone-level audio transcriptions, but the pronunciation dictionaries used by it are typically based on standard varieties of US English, leading to errorful outputs for non-standard varieties. We employ a customized pronunciation dictionary with the Montreal Forced Aligner to increase labeling accuracy of the prenasal merger (a.k.a. pin-pen merger) in Southern US English. We allow the aligner to choose between IH (/ɪ/) and EH (/ɛ/) in words where the merger is expected, rather than enforcing a standard, unmerged pronunciation. We examine the tokens reclassified from EH to IH when using the new dictionary, and we use formant values to study the acoustic separation (measured by Pillai scores and Euclidean distances between centroids) between vowel formant clusters. When applied to the Digital Archive of Southern Speech (DASS), we find that the modification increases the separation between the prenasal allophones of IH and EH, and also that the proportion of prenasal EH tokens reclassified to IH is correlated with the original degree of separation between prenasal IH and EH for each DASS speaker. K-means clustering is also used to show the modification yields more accurate phonetic transcriptions, measured by increased precision and recall.

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 distilled prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.

metaresearch head score (Codex)0.001
metaresearch head score (Gemma)0.000
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Simulation or modeling · Consensus signal: none
GenreCandidate signal: Methods · Consensus signal: Methods
Teacher disagreement score0.963
Threshold uncertainty score0.255

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0010.000
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.001
Bibliometrics0.0000.000
Science and technology studies0.0000.000
Scholarly communication0.0000.000
Open science0.0010.000
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0000.000

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.

Opus teacher head0.019
GPT teacher head0.260
Teacher spread0.241 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one teacher head, not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designSimulation or modeling
Domainnot available
GenreMethods

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".

Quick stats

Citations1
Published2019
Admission routes1
Has abstractyes

Explore more

Same venueThe Journal of the Acoustical Society of AmericaSame topicSpeech Recognition and SynthesisFrench-language works237,207