快速且鲁棒的设备端说话人日志：步长加速管道的相对最小聚类大小
Fast and Robust On-Device Speaker Diarization: Relative Minimum Cluster Size for Stride-Accelerated Pipelines
AI总结：
针对设备端说话人日志的推理成本问题，提出相对最小聚类大小（mcs=round(f*n), f=0.01）以自适应嵌入预算，在保持AMI上DER不变的同时，将VoxConverse的DER从0.113恢复至0.079，加速比达12.2倍。
摘要：
诸如会议转录和语音助手等语音应用将受益于设备端说话人日志，但实际采用受限于推理成本。我们研究了基于Pyannote 3.1的管道在消费级硬件（RTX 5070 Ti GPU和Apple M4笔记本）上能在多大程度上加速，同时保持说话人日志错误率（DER）。一个简单的方案：更粗的分割步长和逐块嵌入，在AMI上实现了多倍加速且DER不变，但在野外数据上急剧退化：在VoxConverse上，DER从0.075上升到0.113。我们将失败归因于聚类阶段说话人计数不足，这是由于固定的最小聚类大小与每个说话人嵌入数量减少相互作用所致。我们提出相对最小聚类大小，mcs = round(f * n)，其中f = 0.01，它自适应于每个录音的嵌入预算。单个f值将VoxConverse DER恢复至0.079（约恢复丢失准确率的89%），同时保持AMI不变，加速后的管道在AMI（MPS）上相对于我们的CAM++基线达到12.2倍加速。
英文摘要：
Speech applications such as meeting transcription and voice agents would benefit from on-device speaker diarization, but practical adoption is limited by inference cost. We study how far a Pyannote 3.1-based pipeline can be accelerated on consumer hardware (an RTX 5070 Ti GPU and an Apple M4 laptop) while preserving diarization error rate (DER). A simple recipe: coarser segmentation stride and per-chunk embedding, yields multi-fold speedups and is DER-neutral on AMI, but degrades sharply on in-the-wild data: on VoxConverse, DER rises from 0.075 to 0.113. We trace the failure to speaker under-counting in the clustering stage, caused by a fixed minimum cluster size interacting with the reduced number of embeddings per speaker. We propose a relative minimum cluster size, mcs = round(f * n) with f = 0.01, which adapts to the embedding budget per recording. A single value of f recovers VoxConverse DER to 0.079 (about 89% of the lost accuracy) while keeping AMI flat, and the accelerated pipeline reaches up to 12.2x speedup on AMI (MPS) over our CAM++ baseline.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
