Multilingual and Fully Non-Autoregressive ASR with Large Language Model Fusion: A Comprehensive Study

2024/01/23 by Huang, W. Ronny, Allauzen, Cyril, Chen, Tongzhou +7 · 3 citations
#Audio and Speech Processing (eess.AS) #Computation and Language (cs.CL) #FOS: Computer and information sciences #FOS: Electrical engineering #Sound (cs.SD) #electronic engineering #information engineering

paper · doi:10.48550/arxiv.2401.12789

Abstract

In the era of large models, the autoregressive nature of decoding often results in latency serving as a significant bottleneck. We propose a non-autoregressive LM-fused ASR system that effectively leverages the parallelization capabilities of accelerator hardware. Our approach combines the Universal Speech Model (USM) and the PaLM 2 language model in per-segment scoring mode, achieving an average relative WER improvement across all languages of 10.8% on FLEURS and 3.6% on YouTube captioning. Furthermore, our comprehensive ablation study analyzes key parameters such as LLM size, context length, vocabulary size, fusion methodology. For instance, we explore the impact of LLM size ranging from 128M to 340B parameters on ASR performance. This study provides valuable insights into the factors influencing the effectiveness of practical large-scale LM-fused speech recognition systems.

Cited by

• Towards Effective and Efficient Non-autoregressive decoders for Conformer and LLM-based ASR using Block-based Attention Mask
• Drax: Speech Recognition with Discrete Flow Matching
• TokenVerse++: Towards Flexible Multitask Learning with Dynamic Task Activation

Related

• Delayed Fusion: Integrating Large Language Models into First-Pass Decoding in End-to-end Speech Recognition
• Anatomy of Industrial Scale Multilingual ASR
• An Embarrassingly Simple Approach for LLM with Strong ASR Capacity
• Efficient Scaling for LLM-based ASR
• Large Language Model Guided Decoding for Self-Supervised Speech Recognition