Primary sourcearXiv cs.AI
View sourceEvent ·Multi-source
Acoustic-to-Text KV压缩:让全双工语音模型长对话内存降64.6%
论文提出 acoustic-to-text KV compression,在全双工语音语言模型的 listening-time slack 区间内,通过转录旁路把语音转成紧凑的文本记忆。推理时 KV 缓存超过预算即淘汰旧声学状态,保留转录与近期声学上下文。旁路用 LoRA 训练,并以知识蒸馏保持原始模型的听说行为。在 MiniCPM-o 4.5 上的实现使十分钟 LongSpeech 会话的峰值流式 KV 缓存降低 64.6%,转录、时序问答和摘要指标同时优于基线,Full-Duplex-Bench 上的停顿、轮次切换和打断表现持平。
Current read
全双工语音模型聊久了内存会爆,这篇用转录旁路压缩KV,十分钟会话峰值缓存直接砍掉64.6%,性能还不掉。
3 sources41° AI heatLast updated 2026/9/25 13:04:09
Evidence
3 sourcesConflict check
Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.