SARGes: Semantically Aligned Reliable Gesture Generation via Intent Chain

Co-speech gesture generation enhances human-computer interaction realism through speech-synchronized gesture synthesis. However, generating semantically meaningful gestures remains a challenging problem. We propose SARGes, a novel framework that leverages large language models (LLMs) to construct an intent chain for parsing speech content and generating reliable semantic gesture labels, which subsequently guide the synthesis of meaningful co-speech gestures. First, we constructed a comprehensive co-speech gesture ethogram and developed an LLM-based intent chain reasoning mechanism that systematically parses and decomposes gesture semantics into structured inference steps following ethogram criteria, effectively guiding LLMs to parse context-aware gesture labels. Subsequently, we constructed a text-to-gesture label dataset and trained a lightweight gesture label generation model, which then guides the generation of credible and semantically coherent co-speech gestures. Experimental results show that SARGes achieves gesture labeling performance comparable to GPT-4 in intent interpretation, with efficient single-pass inference (0.4 seconds), and significantly improves the semantic expressiveness of gesture generation.