VoxMem benchmark: no audio LLM exceeds 40% accuracy on multi-session voice memory
新智元 · wechat · 2026-10-09
A University of Melbourne and UNSW team introduces VoxMem, a multi-session voice memory benchmark spanning 15 'acoustic evidence × memory operation' combinations (799 questions, 177 hours of audio), with questions held constant across 8K–64K token histories. Results: none of 15 audio LLMs exceeds 40% overall accuracy at 32K; tracking tone changes (3.4%) and background-sound changes (1.2%) is near zero. Models remember what was said far better than who said it or how; error analysis shows failures are mostly in retrieval and binding, not reasoning. Paper: arxiv.org/abs/2609.32607
More from Research
- MathEconLib open-sourced: Axiom Math and top economists bring machine-checked proofs to economics in Lean 4 — soumitrashukla9 · 2026-10-23
- Ken Ono invites the community to build MathEcon on the new Lean 4 economics library — soumitrashukla9 · 2026-10-23
- IFM open-sources RL360: 8,192 concurrent sandboxes and ~1.3M rollouts for agentic RL — rupspace · 2026-10-23
- Science lifts expression of concern on 2024 AI debunking-conspiracies paper; conclusions unchanged — DG_Rand · 2026-10-23
- Embodied First Exam benchmarks coding agents on 1,090 tasks across 44 robots — ManlingLi_ · 2026-10-23
- Anima Anandkumar on neural operators: AI weather models beat supercomputers — AnimaAnandkumar · 2026-10-23