Graduate student Siyu Liang has just won the Senior Area Chair Highlight Award at the Empirical Methods in Natural Language Processing 2025 conference!
The paper, “Beyond WER; Probing Whisper’s Sub-token Decoder Across Diverse Language Resource Levels,” authored by Siyu Liang, Nicholas Ballier, Gina-Anne Levow, and Richard Wright, offers new insight into sub-token level disparities in large speech language models across languages differing in resource level and linguistic typology. Whisper is a Large Language Model introduced by OpenAI in 2022. From the publication, "This paper introduces a fine-grained analysis of Whisper's multilingual decoder, examining its sub-token hypotheses during transcription across languages with various resource levels."
"Beyond WER" has received an SAC Highlight Award at the Empirical Methods in Natural Language Processing (EMNLP) 2025 conference, which was held November 4th-9th in Suzhou, China.
Their paper was funded by UW’s Global Innovation Fund. Dr. Nicolas Ballier is affiliated with Université Paris Cité (UPCité), which is the partner institution for the UW Global Innovation Fund. You can find out more about the fund here: Global Innovation Fund – Office of Global Affairs
The paper is hosted here: Beyond WER: Probing Whisper’s Sub‐token Decoder Across Diverse Language Resource Levels - ACL Anthology