Combining X-Vectors and Bayesian Batch Active Learning: Two-Stage Active Learning Pipeline for Speech Recognition

This paper introduces a novel two-stage active learning (AL) pipeline for automatic speech recognition (ASR), combining unsupervised and supervised AL methods. The first stage utilizes unsupervised AL by using x-vectors clustering for diverse sample selection from unlabeled speech data, thus establi...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:IEEE Transactions on Audio, Speech and Language Processing Ročník 33; s. 1862 - 1876
Hlavní autoři: Kundacina, Ognjen, Vincan, Vladimir, Miskovic, Dragisa
Médium: Journal Article
Jazyk:angličtina
Vydáno: IEEE 2025
Témata:
Active learning
batch active learning
Bayes methods
Bayesian active learning
Data models
Deep learning
Labeling
Pipelines
Recording
speech recognition
Training
Tuning
Uncertainty
x-vectors
ISSN:2998-4173, 2998-4173
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:This paper introduces a novel two-stage active learning (AL) pipeline for automatic speech recognition (ASR), combining unsupervised and supervised AL methods. The first stage utilizes unsupervised AL by using x-vectors clustering for diverse sample selection from unlabeled speech data, thus establishing a robust initial dataset for the subsequent supervised AL. The second stage incorporates a supervised AL strategy, with a batch AL method specifically developed for ASR, aimed at selecting diverse and informative batches of samples. Here, sample diversity is also achieved using x-vectors clustering, while the most informative samples are identified using a Bayesian AL method tailored for ASR with an adaptation of Monte Carlo dropout to approximate Bayesian inference. This approach enables precise uncertainty estimation, thereby enhancing ASR model training with significantly reduced data requirements. Our method has shown superior performance compared to competing methods on homogeneous, heterogeneous, and OOD test sets, demonstrating that strategic sample selection and innovative Bayesian modeling can substantially optimize both labeling effort and data utilization in deep learning-based ASR applications.
ISSN:2998-4173
2998-4173
DOI:10.1109/TASLPRO.2025.3565216