Investigating ASR Error Correction with Large Language Model and Multilingual 1-best Hypotheses | Synapse