ViSQA: A benchmark dataset and baseline models for Vietnamese spoken question answering.

A notable trend across all models is the continued difficulty with “Why” and “Others” question types, where accuracies remain significantly lower than other categories.

also in 2,380 other papers