IKFST: IOO and KOO Algorithms for Accelerated and Precise WFST-based End-to-End Automatic Speech Recognition

2026-01-01

arXiv.org

AI Summary

This work addresses the inefficiency of weighted finite-state transducer (WFST) decoding in connectionist temporal classification (CTC) models, which relies on frame-by-frame autoregressive search. The study is the first to explicitly distinguish the structural roles of blank frames—carrying positional information—and non-blank frames—conveying semantic content—in CTC outputs. Building on this insight, the authors propose two efficient decoding algorithms, IOO and KOO, that significantly accelerate WFST-based decoding while preserving end-to-end speech recognition accuracy. Experimental results on AISHELL-1, LibriSpeech, and a large-scale internal dataset demonstrate that the proposed approach maintains state-of-the-art recognition performance while substantially reducing decoding latency.

Technology Category

Cognitive Modeling & Cognitive Systems: Neural Spike CodingMachine Learning: Deep Generative Models & AutoencodersPlanning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphsSemantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semanticsSearch and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

End-to-end automatic speech recognition has become the dominant paradigm in both academia and industry. To enhance recognition performance, the Weighted Finite-State Transducer (WFST) is widely adopted to integrate acoustic and language models through static graph composition, providing robust decoding and effective error correction. However, WFST decoding relies on a frame-by-frame autoregressive search over CTC posterior probabilities, which severely limits inference efficiency. Motivated by establishing a more principled compatibility between WFST decoding and CTC modeling, we systematically study the two fundamental components of CTC outputs, namely blank and non-blank frames, and identify a key insight: blank frames primarily encode positional information, while non-blank frames carry semantic content. Building on this observation, we introduce Keep-Only-One and Insert-Only-One, two decoding algorithms that explicitly exploit the structural roles of blank and non-blank frames to achieve significantly faster WFST-based inference without compromising recognition accuracy. Experiments on large-scale in-house, AISHELL-1, and LibriSpeech datasets demonstrate state-of-the-art recognition accuracy with substantially reduced decoding latency, enabling truly efficient and high-performance WFST decoding in modern speech recognition systems.

Problem

• WFST
• end-to-end ASR
• CTC decoding
• inference efficiency
• autoregressive search

Innovation

• WFST decoding
• CTC modeling
• blank/non-blank frame analysis
• efficient inference
• end-to-end ASR