PRoDeliberation: Parallel Robust Deliberation for End-to-End Spoken Language Understanding

TL;DR

PRoDeliberation introduces a non-autoregressive, parallel decoding approach for spoken language understanding that significantly reduces latency while maintaining robustness against speech recognition errors.

Contribution

It presents a novel CTC-based decoding and denoising training method for non-autoregressive SLU models, improving speed and robustness over previous autoregressive systems.

Findings

Achieves 2-10x latency reduction compared to autoregressive models.

Retains ability to correct ASR mistranscriptions.

Overcomes limitations of small ASR devices through denoising training.

Abstract

Spoken Language Understanding (SLU) is a critical component of voice assistants; it consists of converting speech to semantic parses for task execution. Previous works have explored end-to-end models to improve the quality and robustness of SLU models with Deliberation, however these models have remained autoregressive, resulting in higher latencies. In this work we introduce PRoDeliberation, a novel method leveraging a Connectionist Temporal Classification-based decoding strategy as well as a denoising objective to train robust non-autoregressive deliberation models. We show that PRoDeliberation achieves the latency reduction of parallel decoding (2-10x improvement over autoregressive models) while retaining the ability to correct Automatic Speech Recognition (ASR) mistranscriptions of autoregressive deliberation systems. We further show that the design of the denoising training allows PRoDeliberation to overcome the limitations of small ASR devices, and we provide analysis on the necessity of each component of the system.