Speech Utterance
Speech utterance research focuses on understanding how spoken language is produced, interpreted, and utilized in various contexts, aiming to improve human-computer interaction and communication technologies. Current research emphasizes leveraging large language models (LLMs) for tasks like real-time translation, generating realistic dialogues, and improving speech recognition, often incorporating techniques like fine-tuning BERT, phoneme monotonic alignment, and asynchronous multi-LLM frameworks. This field is crucial for advancing applications such as virtual assistants, multilingual communication tools, and assistive technologies for individuals with communication impairments, while also addressing challenges like bias in sentiment analysis and ensuring the safety and ethical use of LLMs in dialogue systems.
Papers
The Goldilocks of Pragmatic Understanding: Fine-Tuning Strategy Matters for Implicature Resolution by LLMs
Laura Ruis, Akbir Khan, Stella Biderman, Sara Hooker, Tim Rocktäschel, Edward Grefenstette
Question-Interlocutor Scope Realized Graph Modeling over Key Utterances for Dialogue Reading Comprehension
Jiangnan Li, Mo Yu, Fandong Meng, Zheng Lin, Peng Fu, Weiping Wang, Jie Zhou