Filler Word

Filler words, such as "uh" and "um," are common in spontaneous speech and are the subject of ongoing research focusing on their automatic detection and classification. Current research employs various machine learning models, including deep learning architectures like neural semi-Markov conditional random fields and Voice Activity Projection models, to improve the accuracy of filler word identification, often leveraging automatic speech recognition (ASR) systems but also exploring transcription-free approaches. This research is significant for improving the quality of speech processing applications, such as automated transcription and media editing, and for advancing our understanding of spoken language processing and human communication. Furthermore, research is exploring the role of filler words in dialogue generation and their impact on conversational flow.