Automatic Chord Recognition (ACR) is constrained by the scarcity of aligned chord labels, as well-aligned annotations are costly to acquire. At the same time, open-weight pre-trained models are currently more accessible than their proprietary training data. In this work, we present a two-stage training pipeline that leverages pre-trained models together with unlabeled audio. The proposed method decouples training into two stages. In the first stage, we use a pre-trained BTC model as a teacher to generate pseudo-labels for over 1,000 hours of diverse unlabeled audio and train a student model solely on these pseudo-labels. In the second stage, the student is continually trained on ground-truth labels as they become available, with selective knowledge distillation (KD) from the teacher applied as a regularizer to prevent catastrophic forgetting of the representations learned in the first stage. In our experiments, two models (BTC, 2E1D) were used as students. In stage 1, using only pseudo-labels, the BTC student achieves over 98% of the teacher's performance, while the 2E1D model achieves about 96% across seven standard mir_eval metrics. After a single training run for both students in stage 2, the resulting BTC student model surpasses the traditional supervised learning baseline by 2.5% and the original pre-trained teacher model by 1.55% on average across all metrics. And the resulting 2E1D student model improves from the traditional supervised learning baseline by 3.79% on average and achieves almost the same performance as the teacher. Both cases show the large gains on rare chord qualities.

翻译：自动和弦识别（ACR）受限于对齐和弦标签的稀缺性，因为获取良好对齐的标注成本高昂。与此同时，开放权重的预训练模型目前比其专有训练数据更易获得。在本工作中，我们提出了一种两阶段训练流程，该流程利用预训练模型与未标注音频数据。所提方法将训练解耦为两个阶段。在第一阶段，我们使用预训练的BTC模型作为教师，为超过1000小时多样化的未标注音频生成伪标注，并仅在这些伪标注上训练学生模型。在第二阶段，当真实标注可用时，学生模型继续在其上进行训练，并采用选择性知识蒸馏（KD）作为正则化方法，以防止第一阶段学习到的表征发生灾难性遗忘。在我们的实验中，使用了两种模型（BTC、2E1D）作为学生模型。在第一阶段，仅使用伪标注时，BTC学生模型达到了教师模型性能的98%以上，而2E1D模型在七项标准mir_eval指标上均达到约96%。在第二阶段对两个学生模型进行单次训练后，所得的BTC学生模型在所有指标上平均超越传统监督学习基线2.5%，并平均超越原始预训练教师模型1.55%。而所得的2E1D学生模型则平均从传统监督学习基线提升了3.79%，并达到了与教师模型几乎相同的性能。两种情形均显示出在罕见和弦性质识别上的显著提升。