论文解读
终点相同不等于负担相同：语音客服审计为何要先过验证门
该文把语音客服审计对象定为固定服务事实下的完整服务事件，用七道验证门和六类指标区分最终结果与过程负担，并以全合成退款实例说明方法，全程不报告真系统比较结果。
语音对话系统
论文解读
边说话边动：把全身动作做成对话模型的原生输出
针对先说话后做动作的级联需要两次推理且无法联合优化的问题，Motion-Omni 让动作直接从产生语音的隐状态生成，并在 422856 对教师伪标签与 SwDA-500 对照下以 RTF=0.78 达到与同音频教师级联接近的动作质量，代价是动作质量仍受教师分布约束。
论文解读
语音不只是文字的喇叭：温暖介导伤害为何必须同时审听觉与文本
在同一模型上以 7 轮 WHO 脚本对比音频与纯文本，音频在诱发轮出现更短更快更低更轻且不更温暖的韵律漂移，文本转录审计会漏检而高风险自伤脚本的模态差异最集中。
论文解读
语音上下文不能只存文字：用有界编排同时管住说什么、怎么说和何时打断
llmovoice 把语音交互的上下文拆成语义、副语言与环境三状态，用 VoicePage/VoiceThread 组织记忆并在预算内投影到上下文窗口，再让服务模型联合推理生成运行时指令，在丢包下把误打断率从 46.0% 降至 0.9%、每轮成本最高降低 24.9 倍，代价是编排调用时约 790 ms 的额外决策延迟与摘要压缩带来的细节丢失。
论文解读
听见的对话如何被核查：TRILOGUE 把多轮声称钉回源文章与配对音频
针对播客式多轮对话中声称分散与 ASR 噪声的核查难题，TRILOGUE 以源文章可追溯的英俄哈三语对话与配对音频构建受控基准，显示仅看声称的验证最难、加入源证据后大幅回升，而哈萨克语的识别与合成瓶颈仍是主要代价。
论文解读
Reference-Based Prosody and Rhythm Evaluation for Spoken Dialogue Systems
语音对话系统 | 4.7/10
论文解读
IHBench: Evaluating Post-Interruption Recovery in Voice Agents with Structured Workflows
语音对话系统 | 7.5/10
论文解读
MuVAP: Multimodal Multiparty Voice Activity Projection for Turn-taking Prediction in the Wild
语音对话系统 | 7.8/10
论文解读
Multi-Faceted Interactivity Alignment in Full-Duplex Speech Models
语音对话系统 | 9.3/10
论文解读
Proactive for Uncertainty: Cause-Aware Error Diagnosis and Interactive Clarification for Spoken Dialogue Systems
语音识别 | 9.6/10
论文解读
DuplexSLA: A Full-Duplex Spoken Language Model with Synchronized Speech, Language, and Action
语音对话系统 | 7.8/10
论文解读
Synchronization and Turn-Taking in Full-Duplex Speech Dialogue Models
语音对话系统 | 7.6/10
论文解读
Thinking-while-speaking: A Controlled, Interleaved Reasoning Method for Real-Time Speech Generation
语音对话系统 | 6.9/10
论文解读
From Text to Voice: A Reproducible and Verifiable Framework for Evaluating Tool Calling LLM Agents
模型评估 | 6.3/10
论文解读
EVA-Bench: A New End-to-end Framework for Evaluating Voice Agents
语音对话系统 | 8.0/10
论文解读
How Should LLMs Listen While Speaking? A Study of User-Stream Routing in Full-Duplex Spoken Dialogue
语音对话系统 | 6.0/10
论文解读
MIST: Multimodal Interactive Speech-based Tool-calling Conversational Assistants for Smart Homes
MIST: Multimodal Interactive Speech-based Tool-calling Conversational Assistants for Smart Homes
论文解读
MiniMind-O Technical Report: An Open Small-Scale Speech-Native Omni Model
语音对话系统 | 7.5/10
论文解读
MiniMind-O Technical Report: An Open Small-Scale Speech-Native Omni Model
语音对话系统 | 7.5/10
论文解读
Can Speech LLMs Think while Listening?
语音对话系统 | 7.5/10