Reproduce “What Do Humans Hear When Interacting? Experiments on Selective Listening for Evaluating ASR of Spoken Dialogue Systems” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Kiyotada Mori, Seiya Kawano, Chaoran Liu, Carlos Toshinori Ishi, Angel Fernando Garcia Contreras, Koichiro Yoshino
The authors examine whether speech recognition for dialogue should weight words by their relevance to a reply. Participants responded to Japanese speech, then recalled it; POS-based regression weights from these transcripts define candidate Human-Weighted WER. In an audio-plus-visual comparison, human transcripts had higher WER but higher semantic similarity than Whisper base. Post-response recall limits interpretation, and H-WWER remains exploratory.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.