通过时间池化提升语音验证模型效率，实现更高精度与更低计算量的平衡。
ReDimNet2: Scaling Speaker Verification via Time-Pooled Dimension Reshaping
- 在1D处理路径中引入时间维度池化，保持特征空间特性的同时增强通道扩展能力。
- 模型参数量1.1M至12.3M，计算量0.33至13 GMACS，VoxCeleb1上达到0.287% EER。
- 适合部署在资源受限设备上的高精度语音验证系统，尤其关注效率与性能平衡场景。
我们提出ReDimNet2，一种改进的神经网络架构，用于提取话语级说话人表征，基于ReDimNet的维度重塑框架。核心改进是在1D处理路径中引入时间维度的池化操作。该操作保持了1D特征空间的本质，因为无论时间分辨率如何，1D特征始终是2D特征的重塑版本；同时实现了通道维度的更激进扩展，而计算开销未成比例增加。我们设计了一组七种模型配置（B0-B6），参数量从1.1M到12.3M，计算量从0.33到13 GMACS。在VoxCeleb1基准测试中，ReDimNet2在每个规模点上均优于ReDimNet，实现12.3M参数和13 GMACS下的0.287% EER。
原文摘要 · Abstract (English)
We present ReDimNet2, an improved neural network architecture for extracting utterance-level speaker representations that builds upon the ReDimNet dimension-reshaping framework. The key modification in ReDimNet2 is the introduction of pooling over the time dimension within the 1D processing pathway. This operation preserves the nature of the 1D feature space, since 1D features remain a reshaped version of 2D features regardless of temporal resolution, while enabling significantly more aggressive scaling of the channel dimension without proportional compute increase. We introduce a family of seven model configurations (B0-B6) ranging from 1.1M to 12.3M parameters and 0.33 to 13 GMACS. Experimental results on VoxCeleb1 benchmarks demonstrate that ReDimNet2 improves the Pareto front of computational cost versus accuracy at every scale point compared to ReDimNet, achieving 0.287% EER on Vox1-O with 12.3M parameters and 13 GMACS.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。