通过缓解信噪比数据不平衡改进基于域自适应的语音增强的重混过程
Improved Remixing Process for Domain Adaptation-Based Speech Enhancement by Mitigating Data Imbalance in Signal-to-Noise Ratio
AI总结:
针对基于域自适应的语音增强中伪数据信噪比不平衡导致性能欠佳的问题,提出采用课程学习覆盖更广信噪比范围,经CHiME-7 UDASE任务数据集验证可提升低占比数据的增强性能。
摘要:
RemixIT和Remixed2Remixed是基于域自适应的语音增强(DASE)方法,它们采用全监督训练的教师模型,通过重混教师模型的输出生成伪配对数据。用于增强真实世界录制信号的学生模型,在无真实标注的情况下使用这些伪配对数据训练。由于带噪信号是在自然环境中录制的,数据集不可避免地存在部分声学属性的数据不平衡问题,导致代表性不足的数据对应的性能欠佳。在监督学习中本应平衡的信噪比(SNR)就是一个典型例子。本文利用CHiME-7 UDASE任务的数据集,提供了伪数据的SNR对模型性能有显著影响的实证证据,凸显了SNR平衡在DASE中的重要性。此外,我们提出采用课程学习来覆盖更广泛的SNR范围,以提升代表性不足数据的性能。
英文摘要:
RemixIT and Remixed2Remixed are domain adaptation-based speech enhancement (DASE) methods that use a teacher model trained in full supervision to generate pseudo-paired data by remixing the outputs of the teacher model. The student model for enhancing real-world recorded signals is trained using the pseudo-paired data without ground truth. Since the noisy signals are recorded in natural environments, the dataset inevitably suffers data imbalance in some acoustic properties, leading to subpar performance for the underrepresented data. The signal-to-noise ratio (SNR), inherently balanced in supervised learning, is a prime example. In this paper, we provide empirical evidence that the SNR of pseudo data has a significant impact on model performance using the dataset of the CHiME-7 UDASE task, highlighting the importance of balanced SNR in DASE. Furthermore, we propose adopting curriculum learning to encompass a broad range of SNRs to boost performance for underrepresented data.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。