提出多层级语音知识蒸馏法，提升说话人验证准确率5%。
Integrated Multi-Level Knowledge Distillation for Enhanced Speaker Verification
- 融合不同时间尺度的语音特征，通过梯度敏感表示传递知识
- 在VoxCeleb1数据集上使等错误率降低5%
- 特别适合对实时性与精度有要求的说话人验证场景
知识蒸馏（KD）在音频任务中广泛应用，如说话人验证（SV），通过将大型教师模型的知识迁移到小型学生模型以提高效率和可部署性。现有SV领域的KD方法多沿用图像处理中的范式，主要关注预测概率和隐藏表示的近似，但忽略了语音信号的多层次时序特性。本文提出一种新型知识蒸馏方法——集成多层级知识蒸馏（IML-KD），旨在从教师模型中迁移语音在不同时间尺度上的特征知识。在IML-KD中，教师模型的时序上下文信息被整合进基于梯度的输入敏感表示，该表示来自不同持续时间的语音片段，学生模型则通过多层级对齐学习推断这些表示。我们在VoxCeleb1数据集上进行了说话人验证实验。结果表明，IML-KD显著提升了知识蒸馏性能，等错误率（EER）降低了5%。
原文摘要 · Abstract (English)
Knowledge distillation (KD) is widely used in audio tasks, such as speaker verification (SV), by transferring knowledge from a well-trained large model (the teacher) to a smaller, more compact model (the student) for efficiency and portability. Existing KD methods for SV often mirror those used in image processing, focusing on approximating predicted probabilities and hidden representations. However, these methods fail to account for the multi-level temporal properties of speech audio. In this paper, we propose a novel KD method, i.e., Integrated Multi-level Knowledge Distillation (IML-KD), to transfer knowledge of various temporal-scale features of speech from a teacher model to a student model. In the IML-KD, temporal context information from the teacher model is integrated into novel Integrated Gradient-based input-sensitive representations from speech segments with various durations, and the student model is trained to infer these representations with multi-level alignment for the output. We conduct SV experiments on the VoxCeleb1 dataset to evaluate the proposed method. Experimental results demonstrate that IML-KD significantly enhances KD performance, reducing the Equal Error Rate (EER) by 5%.
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。