TurboBias: Universal ASR Context-Biasing powered by GPU-accelerated Phrase-Boosting Tree

TL;DR

This paper introduces TurboBias, a GPU-accelerated, universal context-biasing framework for ASR that improves accuracy and speed across various model types without requiring additional training.

Contribution

It presents a novel, GPU-accelerated word boosting tree framework supporting all major ASR models, enabling fast, accurate context-biasing with large phrase sets without retraining.

Findings

Outperforms existing open-source biasing methods in accuracy

Maintains high decoding speed with up to 20K key phrases

Supports all major ASR model architectures

Abstract

Recognizing specific key phrases is an essential task for contextualized Automatic Speech Recognition (ASR). However, most existing context-biasing approaches have limitations associated with the necessity of additional model training, significantly slow down the decoding process, or constrain the choice of the ASR system type. This paper proposes a universal ASR context-biasing framework that supports all major types: CTC, Transducers, and Attention Encoder-Decoder models. The framework is based on a GPU-accelerated word boosting tree, which enables it to be used in shallow fusion mode for greedy and beam search decoding without noticeable speed degradation, even with a vast number of key phrases (up to 20K items). The obtained results showed high efficiency of the proposed method, surpassing the considered open-source context-biasing approaches in accuracy and decoding speed. Our context-biasing framework is open-sourced as a part of the NeMo toolkit.