Signals 4 · free daily AI digest

Candor-LR: A Dyadic Conversational Dataset for Audio-Visual Speech Recognition

Paper recorded by Signals 4 on 2026-09-09 in cs.CV. Abstract reproduced from arXiv; link to the original below.

Published 2026-09-09 on arXiv · recorded by Signals 4 on 2026-09-10

Category: cs.CV · 计算机视觉 · first seen 2026-09-10

Abstract

Current audio-visual speech recognition (AVSR) benchmarks, like LRS3, rely heavily on clean, scripted and rehearsed speech. They fail to reflect the complexity of natural conversation, which involves overlapping speech, spontaneous turn-taking, unscripted vocabulary and variable acoustic conditions. To shift the field toward realistic dialogue, we introduce Candor-LR, a conversational benchmark de

Read on arXiv →

#104 most recent of 237 cs.CV papers we have recorded · ↑ newer: Advanced Brain Tissue Imaging with Data-Consistent Diffusion Priors in · ↓ older: Enhanced Deformable Convolution with Center-invariant Offset and Edge-
Cite this page: Candor-LR: A Dyadic Conversational Dataset for Audio-Visual Speech Recognition: the #104 most recent of 237 cs.CV papers we have recorded (as of 2026-09-09). Source: Signals 4 (Signals API) — https://data.jiangzhang.ca/signals4/t/papers/candor-lr-a-dyadic-conversational-dataset-for-audio-visual-speech-recognition.html
Free to quote with attribution to “Signals 4 (Signals API)”. Machine-readable: papers.json
Related: More cs.CV papers · arXiv signals · All papers · Today in AI
Get 4 AI signals a day by email — free.
Subscribe free → See all plans →
Get 4 AI signals a day by email — free
All models · All repos · By company · Daily editions