Modality Dropout for Multimodal Device-Directed Speech Detection Using Verbal and Nonverbal Features
Device-directed speech detection (DDSD) is the binary classification task of distinguishing between queries directed to a voice assistant and parallel ...
