Dysfluent WFST: A Framework for Zero-Shot Speech Dysfluency Transcription and Detection

TL;DR

This paper presents Dysfluent-WFST, a zero-shot speech dysfluency transcription and detection framework that improves accuracy without additional training, aiding clinical assessment of disordered speech.

Contribution

Introduces Dysfluent-WFST, a novel zero-shot decoder that transcribes phonemes and detects dysfluency using existing encoders, outperforming previous methods without extra training.

Findings

Achieves state-of-the-art phonetic error rate and dysfluency detection.

Operates effectively with upstream encoders like WavLM.

Lightweight, interpretable, and improves dysfluency processing.

Abstract

Automatic detection of speech dysfluency aids speech-language pathologists in efficient transcription of disordered speech, enhancing diagnostics and treatment planning. Traditional methods, often limited to classification, provide insufficient clinical insight, and text-independent models misclassify dysfluency, especially in context-dependent cases. This work introduces Dysfluent-WFST, a zero-shot decoder that simultaneously transcribes phonemes and detects dysfluency. Unlike previous models, Dysfluent-WFST operates with upstream encoders like WavLM and requires no additional training. It achieves state-of-the-art performance in both phonetic error rate and dysfluency detection on simulated and real speech data. Our approach is lightweight, interpretable, and effective, demonstrating that explicit modeling of pronunciation behavior in decoding, rather than complex architectures, is key to improving dysfluency processing systems.