Search papers, labs, and topics across Lattice.
This paper introduces COALA, a novel framework that enhances speech-augmented language models (SLMs) by integrating contextual biasing to improve automatic speech recognition (ASR) in complex multi-entity scenarios. By mapping SLM latent representations into a discriminative space, COALA effectively quantifies the matching intensity between audio segments and candidate entities, addressing the challenges posed by context-window limitations and training collapse with multi-target utterances. Experimental results on the LibriSpeech benchmark reveal that COALA outperforms existing methods in contextual biasing performance across varying scales of biasing lists, highlighting its robustness and effectiveness.
COALA outperforms existing methods in contextual biasing for ASR, achieving superior performance even in complex multi-entity scenarios.
Contextual biasing seeks to integrate external knowledge into automatic speech recognition (ASR) systems to accurately recognize domain-specific entities. In this paper, we propose COALA (Contextualized ASR Leveraging Biasing Scoring), a robust framework designed to enhance speech-augmented language models (SLMs) in complex multi-entity scenarios. Considering the inherent context-window limitations of SLMs, identifying relevant target entities from a large-scale biasing list is crucial for effective recognition. To this end, COALA maps SLM latent representations into a specialized discriminative space to quantify the matching intensity between audio segments and candidate entities. Furthermore, we address the training collapse in prior study when handling multi-target utterances-where multiple rare words co-occur. Experimental results on the LibriSpeech benchmark demonstrate that COALA consistently achieves superior contextual biasing performance across various biasing list scales.
