一种少样本无参考的双耳空间音频质量评估方法
By combining a pre-trained mono audio feature extraction and dual attention fusion module with a lightweight prediction network, the problem of efficient deployment of binaural spatial audio quality assessment methods in resource-constrained devices is solved, achieving high-precision and low-resource-consumption audio quality assessment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- NINGBO UNIV
- Filing Date
- 2026-05-14
- Publication Date
- 2026-07-17
AI Technical Summary
Existing binaural spatial audio quality assessment methods rely on reference signals that are difficult to obtain, while mono methods lack spatial feature modeling capabilities. Deep learning-based methods have high requirements for labeled data and computational resources, and their generalization ability is weak in scenarios with few samples, making them difficult to deploy efficiently on resource-constrained devices.
We employ a few-sample, no-reference binaural spatial audio quality assessment method. Through a pre-trained mono audio feature extraction module, combined with a dual-attention fusion module and a lightweight prediction network, we perform feature extraction and quality prediction, including feature decomposition, channel-level calibration, cross-channel spatial dependency modeling, and lightweight encoder design. This method is suitable for resource-constrained scenarios.
It achieves high-precision audio quality assessment under no-reference conditions, reduces computational resource consumption, is suitable for resource-constrained scenarios such as VR/AR devices and in-vehicle terminals, and has good generalization ability and fast convergence speed.
Smart Images
Figure CN122201349B_ABST