Emphasizing a data-centric AI approach, 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.

该研究通过强调数据为中心的人工智能方法，引入一种新的两阶段主动学习（AL）流程，用于自动语音识别（ASR），结合了无监督和监督的 AL 方法。该方法通过使用 x-vectors 聚类从未标记的语音数据中选择样本进行多样化的样本选择，从而建立了对后续监督 AL 的强大的初始数据集。与此同时，该方法还结合了专门为 ASR 开发的一种批量 AL 策略，通过使用 x-vectors 聚类实现样本多样性，并使用适应 Monte Carlo dropout 的贝叶斯 AL 方法识别出最具信息量的样本。该方法通过准确的不确定性估计，显著降低了 ASR 模型训练所需的数据量，相比竞争方法在同质、异质和 OOD 测试集上表现出优异的性能，证明了战略性样本选择和创新的贝叶斯建模在基于深度学习的 ASR 应用中可以极大地优化标记工作和数据利用。