Robust Stuttering Detection via Multi-task and Adversarial Learning
This work addresses the problem of automatic stuttering identification for speech pathologists to track disfluencies in persons who stutter, and it is incremental as it applies existing techniques to a new domain.
The paper tackled stuttering detection by using multi-task and adversarial learning to learn robust features, achieving improvements of up to 10%, 6.78%, and 2% in repetitions, blocks, and interjections over the baseline on the SEP-28k dataset.
By automatic detection and identification of stuttering, speech pathologists can track the progression of disfluencies of persons who stutter (PWS). In this paper, we investigate the impact of multi-task (MTL) and adversarial learning (ADV) to learn robust stutter features. This is the first-ever preliminary study where MTL and ADV have been employed in stuttering identification (SI). We evaluate our system on the SEP-28k stuttering dataset consisting of 20 hours (approx) of data from 385 podcasts. Our methods show promising results and outperform the baseline in various disfluency classes. We achieve up to 10%, 6.78%, and 2% improvement in repetitions, blocks, and interjections respectively over the baseline.