Optimizing Contextual Speech Recognition Using Vector Quantization for Efficient Retrieval

TL;DR

This paper introduces a vector quantization-based approximation for cross-attention in neural speech recognition, enabling efficient use of large biasing catalogues and significantly improving accuracy and computational efficiency.

Contribution

It proposes a novel approximation method for cross-attention using vector quantization, allowing large-scale biasing catalogues to be used efficiently in speech recognition.

Findings

Up to 71% relative error rate reduction in personal entity recognition.

20% reduction in compute time for large biasing lists.

85-95% reduction in memory usage with the proposed method.

Abstract

Neural contextual biasing allows speech recognition models to leverage contextually relevant information, leading to improved transcription accuracy. However, the biasing mechanism is typically based on a cross-attention module between the audio and a catalogue of biasing entries, which means computational complexity can pose severe practical limitations on the size of the biasing catalogue and consequently on accuracy improvements. This work proposes an approximation to cross-attention scoring based on vector quantization and enables compute- and memory-efficient use of large biasing catalogues. We propose to use this technique jointly with a retrieval based contextual biasing approach. First, we use an efficient quantized retrieval module to shortlist biasing entries by grounding them on audio. Then we use retrieved entries for biasing. Since the proposed approach is agnostic to the biasing method, we investigate using full cross-attention, LLM prompting, and a combination of the two. We show that retrieval based shortlisting allows the system to efficiently leverage biasing catalogues of several thousands of entries, resulting in up to 71% relative error rate reduction in personal entity recognition. At the same time, the proposed approximation algorithm reduces compute time by 20% and memory usage by 85-95%, for lists of up to one million entries, when compared to standard dot-product cross-attention.