Acoustic Plectrum Guitar Dataset
收藏资源简介:
本数据集名为‘Acoustic Plectrum Guitar Dataset’,由斯里西瓦萨巴拉米亚纳达尔工程学院创建,专注于自动音乐转录(AMT)任务。数据集包含216个波形文件,每个文件记录了使用声学拨片吉他演奏的音符序列。数据集的创建过程结合了预定义的吉他练习和基于隐马尔可夫模型(HMM)的强制Viterbi对齐技术,确保了音符转录的时间精确性。该数据集主要用于训练机器学习系统,特别是神经网络,以实现对单音吉他独奏表演的自动转录,旨在解决音乐数据集的准确性和效率问题。
Named 'Acoustic Plectrum Guitar Dataset', this dataset was developed by Sri Sivasubramaniya Nadar College of Engineering, targeting the Automatic Music Transcription (AMT) task. It consists of 216 waveform files, each documenting a sequence of musical notes performed on an acoustic plectrum guitar. The dataset creation workflow integrates predefined guitar exercises and a forced Viterbi alignment technique based on the Hidden Markov Model (HMM), ensuring the temporal accuracy of note transcription. This dataset is primarily used for training machine learning systems, particularly neural networks, to achieve automatic transcription of monophonic guitar solo performances, aiming to address the accuracy and efficiency issues in music datasets.




