Title: Importance-Weighted Domain Adaptation for Sound Source Tracking Show Chinese title

Title: 重要性加权的声源跟踪域适应

Abstract: In recent years, deep learning has significantly advanced sound source localization (SSL). However, training such models requires large labeled datasets, and real recordings are costly to annotate in particular if sources move. While synthetic data using simulated room impulse responses (RIRs) and noise offers a practical alternative, models trained on synthetic data suffer from domain shift in real environments. Unsupervised domain adaptation (UDA) can address this by aligning synthetic and real domains without relying on labels from the latter. The few existing UDA approaches however focus on static SSL and do not account for the problem of sound source tracking (SST), which presents two specific domain adaptation challenges. First, variable-length input sequences create mismatches in feature dimensionality across domains. Second, the angular coverages of the synthetic and the real data may not be well aligned either due to partial domain overlap or due to batch size constraints, which we refer to as directional diversity mismatch. To address these, we propose a novel UDA approach tailored for SST based on two key features. We employ the final hidden state of a recurrent neural network as a fixed-dimensional feature representation to handle variable-length sequences. Further, we use importance-weighted adversarial training to tackle directional diversity mismatch by prioritizing synthetic samples similar to the real domain. Experimental results demonstrate that our approach successfully adapts synthetic-trained models to real environments, improving SST performance. Show Chinese abstract

Abstract: 近年来，深度学习在声源定位（SSL）方面取得了显著进展。 然而，训练此类模型需要大量的标注数据，而真实录音尤其昂贵，特别是当声源移动时。 虽然使用模拟房间脉冲响应（RIR）和噪声的合成数据提供了一个实用的替代方案，但仅在合成数据上训练的模型在真实环境中会受到领域偏移的影响。 无监督领域自适应（UDA）可以通过在不依赖后者标签的情况下对齐合成和真实领域来解决这个问题。 然而，现有的少数UDA方法主要关注静态SSL，没有考虑到声源跟踪（SST）的问题，这带来了两个特定的领域自适应挑战。 首先，可变长度输入序列会导致不同领域之间的特征维度不匹配。 其次，由于部分领域重叠或由于批量大小限制，合成数据和真实数据的角度覆盖可能无法很好地对齐，我们将其称为方向多样性不匹配。 为了解决这些问题，我们提出了一种针对SST的新型UDA方法，该方法基于两个关键特性。 我们采用循环神经网络的最终隐藏状态作为固定维度的特征表示，以处理可变长度序列。 此外，我们使用重要性加权对抗训练，通过优先考虑与真实领域相似的合成样本，来解决方向多样性不匹配问题。 实验结果表明，我们的方法成功地将合成训练的模型适应到真实环境中，提高了SST性能。