ICM EEG Dataset: EEG Recordings During Perception of Indian Classical Music Ragas

This dataset contains raw electroencephalography (EEG) recordings from nine untrained participants (7 male, 2 female, ages 23–26) collected during the perception of three Indian Classical Music (ICM) ragas. It was created as part of the study: Das, N. & Chakraborty, M. (2025). “Optimal multimodal feature combination and classifier selection for music-based EEG signal analysis.” Computers in Biology and Medicine, 196, 110696. https://doi.org/10.1016/j.compbiomed.2025.110696 Participants listened to 2‑minute alaap sections of Raga Ahir Bhairav, Bhimpalashri, and Yaman Kalyan (performed by Pandit Ravi Shankar), each presented twice in randomized order. The 17‑minute experimental protocol began with a 2‑minute baseline rest, followed by six music segments interleaved with 30‑second rest periods. Raga names were announced ~5 s before each music segment to reorient attention. Detailed description of the experimental protocol is given in the cited study. EEG was recorded with the RMS Maximus 24 system (19 scalp electrodes, international 10–20 system, reference at nasion, hardware band‑pass 0.5–70 Hz, sampling rate 256 Hz). ECG was acquired simultaneously in a standard Lead I configuration using bipolar electrodes. Participants sat with eyes closed in a dimly lit, sound‑controlled room. All participants provided informed consent to participate in this study, and all identifiable information was anonymized. The dataset is distributed as a single archive containing: One unprocessed CSV file per subject (raw_data/), each with 261,120 × 22 samples (1020 s × 256 Hz). phase_events.csv: start/end times (in seconds) for each rest and music segment. music_order.csv: mapping of generic labels (music_1–music_6) to actual ragas for each participant. Detailed documentation (README.md). The EEG data are raw and unprocessed except for trimming to the exact experimental duration. Citation: If you use this dataset in your research, please cite both this dataset (Zenodo DOI) and the associated paper:Das, N. & Chakraborty, M. (2025). Optimal multimodal feature combination and classifier selection for music-based EEG signal analysis. Computers in Biology and Medicine, 196, 110696. https://doi.org/10.1016/j.compbiomed.2025.110696本数据集收录了9名未接受专业训练的受试者（7名男性、2名女性，年龄区间为23~26岁）在聆听三首印度古典音乐（Indian Classical Music, ICM）拉格时采集的原始脑电（electroencephalography, EEG）信号记录。本数据集源自如下研究： Das, N. & Chakraborty, M. (2025). "Optimal multimodal feature combination and classifier selection for music-based EEG signal analysis." Computers in Biology and Medicine, 196, 110696. https://doi.org/10.1016/j.compbiomed.2025.110696 受试者聆听了由潘迪特·拉维·香卡（Pandit Ravi Shankar）演奏的《拉格·阿伊尔·拜拉夫》（Raga Ahir Bhairav）、《拉格·比姆帕拉什里》（Bhimpalashri）与《拉格·亚曼·卡利安》（Yaman Kalyan）的2分钟阿拉普（alaap）段落，每个段落以随机顺序呈现两次。整个实验流程总时长17分钟，首先为2分钟的基线静息阶段，随后是6段音乐片段与30秒静息阶段交替的实验序列。每段音乐片段开始前约5秒，会播报对应拉格的名称以引导受试者重新集中注意力。实验流程的详细说明请参见上述引用文献。 脑电信号采用RMS Maximus 24系统采集（配备19个头皮电极，遵循国际10-20系统标准，参考电极置于鼻根，硬件带通滤波范围为0.5~70Hz，采样率为256Hz）。同步采用双极电极以标准I导联配置采集心电（electrocardiography, ECG）信号。受试者在光线昏暗、音量受控的实验室内闭眼静坐。所有受试者均已签署知情同意书，且所有可识别个人信息均已完成匿名化处理。 本数据集以单个压缩包形式分发，包含以下内容： - 每位受试者对应一份未处理的CSV文件（存放于raw_data/目录下），每份文件包含261120×22个样本（对应1020秒时长 × 256Hz采样率）。 - phase_events.csv：记录各静息与音乐片段的起始、结束时间（单位：秒）。 - music_order.csv：建立每位受试者的通用标签（music_1~music_6）与实际对应拉格的映射关系。 - 详细说明文档（README.md）。 本数据集的脑电数据为原始未处理状态，仅经过裁剪以匹配实验的精确持续时长。 引用须知：若您在研究中使用本数据集，请同时引用本数据集（Zenodo DOI）及相关研究论文：Das, N. & Chakraborty, M. (2025). Optimal multimodal feature combination and classifier selection for music-based EEG signal analysis. Computers in Biology and Medicine, 196, 110696. https://doi.org/10.1016/j.compbiomed.2025.110696