PAREDA, a multi-accent speech dataset of NLP research discussions, published at SPEAKABLE @ LREC 2026.