Search papers, labs, and topics across Lattice.
Keio University, Japan, The University of Tokyo, Japan
4
0
4
28
Reducing inter-utterance silence from 9.6 seconds to 0.3 seconds transforms the quality of real-time game commentary, making it feel more natural and engaging.
Speaker embeddings from foundation models often misalign with human perceptions of similarity, revealing critical gaps in current AI speech processing.
Skip the emotionless text bottleneck: directly transferring prosody from sign language video to speech unlocks more expressive and natural communication.
Forget fine-tuning: Prompting MLLMs with a dynamic interval-based decoding strategy lets them generate surprisingly human-like, pause-aware real-time game commentary.
