MusicNet-16k + EM for YourMT3

<strong>&lt; UNDER CONSTRUCTION &gt;</strong> <strong>About this version:</strong> This particular variant of the MusicNet dataset has been resampled to a 16 kHz-mono-16-bit-wav format, which makes it more suitable for certain audio processing tasks, particularly those that require lower sampling rates. We redistribute this data as a part of YourMT3 project. The license for redistribution is attached. Moreover, this version of the dataset includes various split options derived from previous works on automatic music transcription as python dictionary (see README.md). Below is a brief description of available split options: <pre><code>MUSICNET_SPLIT_INFO = { 'train_mt3': [], # the first 300 songs are synth dataset, while the remaining 300 songs are acoustic dataset. 'train_mt3_synth' : [], # Note: this is not the synthetic dataset of EM (MIDI Pop 80K) nor pitch-augmented. Just recording of MusicNet MIDI, split by MT3 author's split. But not sure if they used this (maybe not). 'train_mt3_acoustic': [], 'validation_mt3': [1733, 1765, 1790, 1818, 2160, 2198, 2289, 2300, 2308, 2315, 2336, 2466, 2477, 2504, 2611], 'validation_mt3_synth': [1733, 1765, 1790, 1818, 2160, 2198, 2289, 2300, 2308, 2315, 2336, 2466, 2477, 2504, 2611], 'validation_mt3_acoustic': [1733, 1765, 1790, 1818, 2160, 2198, 2289, 2300, 2308, 2315, 2336, 2466, 2477, 2504, 2611], 'test_mt3_acoustic': [1729, 1776, 1813, 1893, 2118, 2186, 2296, 2431, 2432, 2487, 2497, 2501, 2507, 2537, 2621], 'train_thickstun': [], # the first 327 songs are synth dataset, while the remaining 327 songs are acoustic dataset. 'test_thickstun': [1819, 2303, 2382], 'train_mt3_em': [], # 293 tracks. MT3 train set - 7 missing tracks[2194, 2211, 2227, 2230, 2292, 2305, 2310], ours 'validation_mt3_em': [1733, 1765, 1790, 1818, 2160, 2198, 2289, 2300, 2308, 2315, 2336, 2466, 2477, 2504, 2611], # ours 'test_mt3_em': [1729, 1776, 1813, 1893, 2118, 2186, 2296, 2431, 2432, 2487, 2497, 2501, 2507, 2537, 2621], # ours 'train_em_table2' : [], # 317 tracks. Whole set - 7 missing tracks[2194, 2211, 2227, 2230, 2292, 2305, 2310] - 6 test_em 'test_em_table2' : [2191, 2628, 2106, 2298, 1819, 2416], # strings and winds from Cheuk's split, using EM annotations 'test_cheuk_table2' : [2191, 2628, 2106, 2298, 1819, 2416], # strings and winds from Cheuk's split, using Thickstun's annotations }</code></pre> <strong>About MusicNet:</strong> The MusicNet dataset, originally released in 2016 by Thickstun et al., "Learning Features of Music from Scratch". It is a collection of music recordings annotated with labels for various tasks, such as automatic music transcription, instrument recognition, and genre classification. The original dataset contains over 330 hours of audio, sourced from various public domain recordings of classical music, and is labeled with instrument activations and note-wise annotations. <strong>About MusicNet EM:</strong> MusicNetEM are refined labels for the MusicNet dataset, in the form of MIDI files. They are aligned with the recordings, with onset timing within 32ms. They were created using an EM process, similar to the one described in the Ben Maman and Amit H. Bermano, "Unaligned Supervision for Automatic Music Transcription in The Wild". Their split (Table 2 of this paper) derived from another paper, Kin Wai Cheuk et al., "ReconVAT: A Semi-Supervised Automatic Music Transcription Framework for Low-Resource Real-World Data". <strong>License:</strong> CC-BY-4.0<strong>&lt; 正在构建中 &gt;</strong> <strong>关于本版本：</strong>本MusicNet数据集的该变体已重采样为16 kHz-单声道-16位WAV格式，更适配部分对采样率要求较低的音频处理任务。我们将此数据作为YourMT3项目的一部分进行重新分发，重新分发许可证已随附。此外，本版本数据集还包含源自此前自动音乐转录（automatic music transcription）相关研究的多种划分方案，以Python字典形式提供（详见README.md）。以下为可用划分方案的简要说明： <pre><code>MUSICNET_SPLIT_INFO = { 'train_mt3': [], # 前300首为合成数据集，剩余300首为原声数据集。 'train_mt3_synth' : [], # 注：此并非EM（MIDI Pop 80K）的合成数据集，也未经过音高增强处理。仅为MusicNet MIDI的录音，采用MT3作者的划分方式，但不确定其是否实际使用了该划分（可能未使用）。 'train_mt3_acoustic': [], 'validation_mt3': [1733, 1765, 1790, 1818, 2160, 2198, 2289, 2300, 2308, 2315, 2336, 2466, 2477, 2504, 2611], 'validation_mt3_synth': [1733, 1765, 1790, 1818, 2160, 2198, 2289, 2300, 2308, 2315, 2336, 2466, 2477, 2504, 2611], 'validation_mt3_acoustic': [1733, 1765, 1790, 1818, 2160, 2198, 2289, 2300, 2308, 2315, 2336, 2466, 2477, 2504, 2611], 'test_mt3_acoustic': [1729, 1776, 1813, 1893, 2118, 2186, 2296, 2431, 2432, 2487, 2497, 2501, 2507, 2537, 2621], 'train_thickstun': [], # 前327首为合成数据集，剩余327首为原声数据集。 'test_thickstun': [1819, 2303, 2382], 'train_mt3_em': [], # 共293条曲目。MT3训练集减去7条缺失曲目[2194, 2211, 2227, 2230, 2292, 2305, 2310]，为本版本划分。 'validation_mt3_em': [1733, 1765, 1790, 1818, 2160, 2198, 2289, 2300, 2308, 2315, 2336, 2466, 2477, 2504, 2611], # 本版本划分 'test_mt3_em': [1729, 1776, 1813, 1893, 2118, 2186, 2296, 2431, 2432, 2487, 2497, 2501, 2507, 2537, 2621], # 本版本划分 'train_em_table2' : [], # 共317条曲目。全量数据集减去7条缺失曲目[2194, 2211, 2227, 2230, 2292, 2305, 2310]，再减去6条test_em曲目。 'test_em_table2' : [2191, 2628, 2106, 2298, 1819, 2416], # 采用EM标注的Cheuk划分中的弦乐与管乐曲目 'test_cheuk_table2' : [2191, 2628, 2106, 2298, 1819, 2416], # 采用Thickstun标注的Cheuk划分中的弦乐与管乐曲目 }</code></pre> <strong>关于MusicNet：</strong>MusicNet数据集由Thickstun等人于2016年在论文《从零开始学习音乐特征》（*Learning Features of Music from Scratch*）中首次发布。该数据集是面向自动音乐转录、乐器识别、流派分类等多种任务构建的音乐录音标注集合，原始数据集包含超过330小时的音频内容，素材来自多首古典音乐的公有领域录音，标注内容涵盖乐器激活状态与逐音符注释。<strong>关于MusicNet EM：</strong>MusicNet EM是MusicNet数据集的精细化标注版本，以MIDI文件形式提供，与原始录音对齐，起始时刻误差不超过32毫秒。其标注通过EM（期望最大化）流程生成，该流程参考自Ben Maman与Amit H. Bermano的论文《野外环境下自动音乐转录的非对齐监督》（*Unaligned Supervision for Automatic Music Transcription in The Wild*）。本数据集采用的划分方案（对应论文表2）源自Kin Wai Cheuk等人的论文《ReconVAT：面向低资源真实场景数据的半监督自动音乐转录框架》（*ReconVAT: A Semi-Supervised Automatic Music Transcription Framework for Low-Resource Real-World Data*）。<strong>许可证：</strong>CC-BY-4.0