UCF's ReVoicer turns voice comments into LLM-assisted peer reviews

Researchers at the University of Central Florida (UCF) have introduced ReVoicer, a prototype system designed to streamline the academic peer-review process. Submitted to arXiv on July 31, 2026, the system allows reviewers to highlight a passage in a paper and then speak or type a train-of-thought comment. A large language model (LLM) cleans up the comment using the surrounding prose as context, tags it by comment type, and anchors it to the specific passage. This lets reviewers capture their immediate reactions without breaking their reading flow.

After the reviewer finishes reading, ReVoicer checks the accumulated notes against a venue-specific rubric, flags coverage gaps (e.g., missing discussion of methodology or related work), and drafts a review composed entirely from the reviewer's own comments. The draft is written to a style guide distilled from the reviewer's past reviews, and the system explicitly introduces no critiques of its own—preserving the reviewer's voice and judgment. The team, including Matt Gottsacker and colleagues, plans to demo at IEEE ISMAR 2026 to gather feedback. The implication: peer review could become faster and more consistent, with LLMs handling formatting and coverage while humans remain the sole source of substantive critique.