Research map / Large language models / Language, safety and interpretability
WearVox: An Egocentric Multichannel Voice Assistant Benchmark for Wearables
Zhaojiang Lin, YONG XU, Kai Sun, Jing Zheng, Yin Huang, Surya Teja Appini, Krish Narang, Renjie Tao, Ishan Kapil Jain, Siddhant Arora, Ruizhi Li, Yiteng Huang and 8 more
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| enable thinking mode (ablation) | Gemini 2.5 Flash · WearVox turn-based micro-avg | accuracy (%): 59.8 -> 71.3 | Improved |
| enable thinking mode (ablation) | Gemini 2.5 Flash · WearVox speech translation | score: 50.3 -> 70.1 | Improved |
| enable thinking mode (ablation) | Gemini 2.5 Flash · WearVox time to first token | TTFT (ms): 1592 -> 5546 | Hurt |
| enable thinking mode (ablation) | Gemini 2.5 Flash · WearVox Search Grounded QA | accuracy (%): 49.0 -> 48.8 | No clear effect |
| add multichannel input (channel 0 plus beamformed) (ablation) | Llama-4-Scout-based SLLM · WearVox Tool Calling | accuracy (%): 58.5 -> 63.9 | Improved |
| add multichannel input (channel 0 plus beamformed) (ablation) | Llama-4-Scout-based SLLM · WearVox Side Talk Rejection | accuracy (%): 85.4 -> 93.9 | Improved |
| add multichannel input (ablation) | Llama-4-Scout-based SLLM · WearVox QA tasks | accuracy (%): 43.3 -> 43.3 (search); 42.5 -> 42.2 (closedbook) | No clear effect |
| add multichannel input (ablation) | Llama-4-Scout-based SLLM · WearVox outdoor noisy | accuracy | Improved |
| test multichannel model on unseen channel pair (ablation) | MC Wearllama · LibriSpeech test-other | WER (%): 8.58 -> 8.21 | Improved |
| evaluate in outdoor settings | most SLLMs · WearVox turn-based | accuracy | Hurt |
Closest papers by meaning
- SING: Spatial Context in Large Language Model for Next-Gen Wearables (ICML 2025 · similarity 0.87)
- Talking Turns: Benchmarking Audio Foundation Models on Turn-Taking Dynamics (ICLR 2025 · similarity 0.85)
- $\tau$-Voice: Benchmarking Full-Duplex Voice Agents on Real-World Domains (ICML 2026 · similarity 0.84)
- WearVQA: A Visual Question Answering Benchmark for Wearables in Egocentric Authentic Real-world scenarios (NeurIPS 2025 · similarity 0.84)
- PhoStream: Benchmarking Real-World Streaming for Omnimodal Assistants in Mobile Scenarios (ICML 2026 · similarity 0.83)
- Benchmarking Egocentric Multimodal Goal Inference for Assistive Wearable Agents (NeurIPS 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
