Wang J.-YJYH-SHING JANG2022-04-252022-04-25202115206149https://www.scopus.com/inward/record.uri?eid=2-s2.0-85115050788&doi=10.1109%2fICASSP39728.2021.9414601&partnerID=40&md5=70bedf26b4c297347e361f22fd12f43ahttps://scholars.lib.ntu.edu.tw/handle/123456789/607419This paper proposes a large-scale dataset for singing transcription, along with some methods for fine-tuning and validating its contents. The dataset is named MIR-ST500, which consists of more than 160,000 notes from 500 pop songs. To create this large-scale dataset, we set some labeling criteria and ask non-experts to label notes. We also perform some adjustments on the annotation to correct minor errors. Finally, to validate the dataset, we train a singing transcription model on MIR-ST500 dataset and evaluate it on various datasets. The result shows that we can certainly construct a better singing transcription model for various purposes using MIR-ST500, which is properly labeled and validated. ? 2021 IEEEAutomatic singing transcriptionDataset preparationDataset validationMusic information retrievalSignal processingFine tuningLarge-scale datasetLarge datasetconference paper10.1109/ICASSP39728.2021.94146012-s2.0-85115050788