ODAQ (Open Dataset of Audio Quality)
收藏资源简介:
ODAQ(开放音频质量数据集)是由弗劳恩霍夫集成电路研究所、鲍尔州立大学和Netflix公司联合创建的一个综合性音频质量研究数据集,旨在探索单声道和双声道音频质量退化问题。该数据集包含多种失真类型和信号,以及相应的主观质量评分,数据量涵盖119个音频条目,采样率为48kHz/24位。数据集通过Mid/Side (MS)和Left/Right (LR)等立体声处理技术生成退化信号,并提供了详细的听觉测试结果。其创建过程包括信号处理、主观评分收集和数据分析,主要用于评估现有和新型客观音频质量指标的效能,特别是在复杂听觉场景下的表现。该数据集的应用领域涵盖音频编码、语音清晰度评估、助听器音频质量优化以及基础心理声学研究,旨在解决音频质量评估中时间和空间维度相互作用的复杂问题。
ODAQ (Open Audio Quality Dataset) is a comprehensive audio quality research dataset jointly developed by Fraunhofer Institute for Integrated Circuits, Ball State University, and Netflix, Inc., aiming to investigate audio quality degradation issues in monophonic and stereophonic audio. This dataset contains various distortion types, audio signals, and corresponding subjective quality scores, with a total of 119 audio entries sampled at 48kHz/24-bit. Degraded audio signals are generated via stereo processing techniques such as Mid/Side (MS) and Left/Right (LR), and detailed auditory test results are provided. Its development pipeline includes signal processing, subjective rating collection, and data analysis, and it is primarily used to evaluate the performance of both existing and novel objective audio quality metrics, particularly their performance in complex auditory scenarios. The applicable fields of this dataset cover audio coding, speech intelligibility assessment, hearing aid audio quality optimization, and basic psychoacoustic research, with the goal of addressing the complex problem of the interaction between temporal and spatial dimensions in audio quality evaluation.

- 1Exploring Perceptual Audio Quality Measurement on Stereo Processing Using the Open Dataset of Audio Quality弗劳恩霍夫集成电路研究所, 鲍尔州立大学, Netflix公司 · 2025年



