Human Latency Conversational Turns for Spoken Avatar Systems

arXiv:2404.16053v1 Announce Type: cross
Abstract: A problem with many current Large Language Model (LLM) driven spoken dialogues is the response time. Some efforts such as Groq address this issue by lightning fast processing of the LLM, but we know from the cognitive psychology literature that in human-to-human dialogue often responses occur prior to the speaker completing their utterance. No amount of delay for LLM processing is acceptable if we wish to maintain human dialogue latencies. In this paper, we discuss …

abstract arxiv avatar cognitive conversational cs.ai cs.cl cs.hc current dialogue groq human issue language language model large language large language model latency lightning literature llm prior processing psychology responses spoken systems type