A unified multichannel far-field speech recognition system: combining neural beamforming with attention based end-to-end model

TL;DR

This paper presents a unified multichannel speech recognition system combining neural beamforming with an attention-based end-to-end model, improving robustness and performance in far-field scenarios through joint training and source direction integration.

Contribution

It introduces a novel framework that integrates neural beamforming with transformer-based speech recognition, jointly trained for enhanced far-field speech recognition performance.

Findings

Achieved 19.26% improvement over baseline

Demonstrated robustness across different microphone array geometries

Validated effectiveness on large in-house datasets

Abstract

Far-field speech recognition is a challenging task that conventionally uses signal processing beamforming to attack noise and interference problem. But the performance has been found usually limited due to heavy reliance on environmental assumption. In this paper, we propose a unified multichannel far-field speech recognition system that combines the neural beamforming and transformer-based Listen, Spell, Attend (LAS) speech recognition system, which extends the end-to-end speech recognition system further to include speech enhancement. Such framework is then jointly trained to optimize the final objective of interest. Specifically, factored complex linear projection (fCLP) has been adopted to form the neural beamforming. Several pooling strategies to combine look directions are then compared in order to find the optimal approach. Moreover, information of the source direction is also integrated in the beamforming to explore the usefulness of source direction as a prior, which is usually available especially in multi-modality scenario. Experiments on different microphone array geometry are conducted to evaluate the robustness against spacing variance of microphone array. Large in-house databases are used to evaluate the effectiveness of the proposed framework and the proposed method achieve 19.26\% improvement when compared with a strong baseline.