SESQA: Semi-Supervised Learning for Speech Quality Assessment
SESQA: Semi-Supervised Learning for Speech Quality Assessment
复制标题
SESQA：用于语音质量评估的半监督学习
DOI：
10.1109/icassp39728.2021.9414052
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Santiago Pascual
中科院分区：
文献类型：
--
作者：
J. Serrà;Jordi Pons;Santiago Pascual
自动语音质量评估是一项重要的横向任务，其进展受到人工注释的缺乏，对未见记录条件的不良泛化以及现有方法缺乏灵活性的阻碍。在这项工作中，我们使用半监督学习方法来解决这些问题，将可用的注释与编程生成的数据相结合，并使用3种不同的优化标准以及5种互补的辅助任务。我们的研究结果表明，这种半监督方法可以将现有方法的误差减少36%以上，同时在可重用特征或辅助输出方面提供额外的好处。改进进一步证实了样本外测试显示有希望的泛化能力。
Automatic speech quality assessment is an important, transversal task whose progress is hampered by the scarcity of human annotations, poor generalization to unseen recording conditions, and a lack of flexibility of existing approaches. In this work, we tackle these problems with a semi-supervised learning approach, combining available annotations with programmatically generated data, and using 3 different optimization criteria together with 5 complementary auxiliary tasks. Our results show that such a semi-supervised approach can cut the error of existing methods by more than 36%, while providing additional benefits in terms of reusable features or auxiliary outputs. Improvement is further corroborated with an out-of-sample test showing promising generalization capabilities.
超宽带语音通信网络的非侵入式语音质量评估
DOI：
10.1109/icassp.2019.8683770
发表时间：
2019
期刊：
ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)
影响因子：
--
作者：
G. Mittag;S. Möller
通讯作者：
S. Möller