用线性注意力提升语音分离效率，速度更快内存更省。
FLASepformer: Efficient Speech Separation with Gated Focused Linear Attention Transformer
- 采用聚焦线性注意力机制，降低计算复杂度。
- 在多个数据集上达到顶尖性能，速度提升1.49至2.29倍。
- 适合需要实时推理的语音处理场景，如会议系统、语音助手。
语音分离面临长序列处理挑战。以往方法通过缩短序列长度并使用Transformer捕捉全局信息，但因注意力模块存在二次时间复杂度，长片段下内存与推理时间仍显著增长。为此，本文提出聚焦线性注意力（Focused Linear Attention），构建线性复杂度的FLASepformer模型。受SepReformer和TF-Locoformer启发，设计两种变体：FLA-SepReformer与FLA-TFLocoformer。此外，引入新门控模块进一步提升性能。在多个数据集上的实验表明，FLASepformer达到当前最优性能，同时内存消耗更低、推理更快。FLA-SepReformer-T/B/L分别实现2.29倍、1.91倍和1.49倍加速，GPU内存占用分别减少15.8%、20.9%和31.9%，验证了模型有效性。
原文摘要 · Abstract (English)
Speech separation always faces the challenge of handling prolonged time sequences. Past methods try to reduce sequence lengths and use the Transformer to capture global information. However, due to the quadratic time complexity of the attention module, memory usage and inference time still increase significantly with longer segments. To tackle this, we introduce Focused Linear Attention and build FLASepformer with linear complexity for efficient speech separation. Inspired by SepReformer and TF-Locoformer, we have two variants: FLA-SepReformer and FLA-TFLocoformer. We also add a new Gated module to improve performance further. Experimental results on various datasets show that FLASepformer matches state-of-the-art performance with less memory consumption and faster inference. FLA-SepReformer-T/B/L increases speed by 2.29x, 1.91x, and 1.49x, with 15.8%, 20.9%, and 31.9% GPU memory usage, proving our model's effectiveness.
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。