SESQA: semi-supervised learning for speech quality assessment
This work addresses the challenge of speech quality assessment for applications requiring accurate and generalizable models, though it appears incremental as it builds on existing semi-supervised methods.
The paper tackled the problem of automatic speech quality assessment, which is hindered by scarce human annotations and poor generalization, by using a semi-supervised learning approach that combined annotations with generated data and multiple optimization criteria, resulting in a more than 36% reduction in error compared to existing methods.
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.