Personalized Audio Quality Preference Prediction Dataset

Dataset for the following paper. Please cite this paper if our dataset is used in your research. Chung-Che Wang, Yu-Chun Lin, Yu-Teng Hsu, and Jyh-Shing Roger Jang, "Personalized Audio Quality Preference Prediction", APSIPA ASC 2023. Here is a brief description of our dataset. For more details, please see our paper. This dataset is designed for personalized audio quality preference prediction. It includes recordings from 5 different mobile phones playing 7 distinct song segments at 2 volume settings. The played audio is recorded by using a binaural microphone and a computer interface. For each volume type, 70 pairs of recorded audio files are formed, where each of the two audio files in one pair corresponds to same song segment played by different mobile phones. For each volume type, each subject is asked to compare at least 14 of the 70 pairs. Subject information, which includes age, gender, and headphone/earphone specifications such as impedance, frequency response range, and sensitivity, are also collected.本数据集对应如下论文，若您的研究中使用本数据集，请引用该文献：Chung-Che Wang、Yu-Chun Lin、Yu-Teng Hsu 与 Jyh-Shing Roger Jang，《个性化音频质量偏好预测》，APSIPA ASC 2023。以下为本数据集的简要说明，更多细节请参阅原论文。本数据集专为个性化音频质量偏好预测任务打造，包含5款不同手机以2种音量设置播放7段不同歌曲片段的录制音频。播放的音频通过双耳麦克风（binaural microphone）与计算机接口（computer interface）录制完成。针对每种音量设置，共生成70组录制音频文件对，每组对中的两段音频对应同一歌曲片段，但由不同手机播放。针对每种音量设置，每位受试者需至少对比70组音频对中的14组。同时收集受试者的相关信息，包括年龄、性别，以及其使用的头戴式耳机/入耳式耳机的参数（如阻抗、频响范围与灵敏度）。

搜集汇总

背景与挑战

背景概述

该数据集专为个性化音频质量偏好预测研究设计，包含5部不同手机播放7段歌曲片段在2种音量设置下的双耳录音，形成音频对供受试者比较。数据集还收集了受试者的人口统计信息和耳机规格，适用于音频质量评估和个性化建模。