Context biasing improves speech recognition on rare names and keywords
BaLEEN: Biasing with Latent Encoded Entities for Context-Aware ASR
Computation and LanguageArtificial Intelligence
Summary
Recognizing unusual words like rare names in spoken language is very hard for computers. The authors introduce BaLEEN, a tool that helps speech recognition systems recognize these tricky words better without changing the main system. BaLEEN uses a language model to capture a list of special words and adds that knowledge into the speech system while it listens. This approach cuts down errors on those words and makes overall transcription more accurate.
What this means in practice
- •For speech recognition developers: Integrate context-aware biasing to reduce errors on rare words without retraining the whole speech model.
- •For voice assistant engineers: Improve recognition of personalized names and domain-specific terms by dynamically injecting context during speech processing.
Authors
Chihiro Taguchi, Yotaro Kubo, Rujikorn Charakorn
Abstract
Transcribing domain-specific entities and rare proper nouns remains a major challenge in automatic speech recognition (ASR). In this paper, we propose BaLEEN (Biasing with Latent Encoded Entities), a lightweight, hypernetwork-based framework for dynamic contextual adaptation without fine-tuning the underlying ASR model. BaLEEN encodes variable-length contextual keywords using a pretrained language model, compresses them into a fixed sequence of latent vectors via a Perceiver bottleneck, and injects context-dependent bias vectors directly into the intermediate encoder representations of the ASR model. Because both the language model and the backbone ASR model remain entirely frozen during training, BaLEEN operates as a plug-and-play adapter that incurs zero computational overhead at inference time when context biases are precomputed. We evaluate our method on a CTC-based ASR model using a Wikipedia-derived corpus with annotated named entities and synthetic speech. Experimental results demonstrate that BaLEEN reduces keyword miss rate by 8.7% on the test set relative to the unbiased baseline while simultaneously improving overall word error rate by 21% and character error rate by 28%.