融合声音方向信息的会议语音分段框架，提升多人对话分离效果。
Spatially-Augmented Sequence-to-Sequence Neural Diarization for Meetings
- 将声源方向估计融入序列到序列模型，增强空间感知能力。
- 离线模式下相对错误率降低7.4%，结合通道注意力提升超19%。
- 适合需要高精度多人语音分离的会议记录与实时转写场景。
本文提出一种空间增强型序列到序列神经说话人分离（SA-S2SND）框架，将由SRP-DNN估计的到达方向（DOA）线索整合至S2SND主干网络中。采用两阶段训练策略：先用单通道音频和DOA特征训练模型，再在多通道输入下以DOA为指导进一步优化。此外，引入模拟DOA生成方案，缓解对匹配多通道语料的依赖。在AliMeeting数据集上，SA-S2SND持续优于S2SND基线，在离线模式下相对错误率（DER）降低7.4%，结合通道注意力时改善超过19%。结果表明，空间线索与跨通道建模高度互补，能在在线与离线设置中均取得优异性能。
原文摘要 · Abstract (English)
This paper proposes a Spatially-Augmented Sequence-to-Sequence Neural Diarization (SA-S2SND) framework, which integrates direction-of-arrival (DOA) cues estimated by SRP-DNN into the S2SND backbone. A two-stage training strategy is adopted: the model is first trained with single-channel audio and DOA features, and then further optimized with multi-channel inputs under DOA guidance. In addition, a simulated DOA generation scheme is introduced to alleviate dependence on matched multi-channel corpora. On the AliMeeting dataset, SA-S2SND consistently outperform the S2SND baseline, achieving a 7.4% relative DER reduction in the offline mode and over 19% improvement when combined with channel attention. These results demonstrate that spatial cues are highly complementary to cross-channel modeling, yielding good performance in both online and offline settings.
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。