HPPNet: Modeling the Harmonic Structure and Pitch Invariance in Piano Transcription. Wei, W., Li, P., Yu, Y., & Li, W. August, 2022. ![HPPNet: Modeling the Harmonic Structure and Pitch Invariance in Piano Transcription [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex While neural network models are making significant progress in piano transcription, they are becoming more resource-consuming due to requiring larger model size and more computing power. In this paper, we attempt to apply more prior about piano to reduce model size and improve the transcription performance. The sound of a piano note contains various overtones, and the pitch of a key does not change over time. To make full use of such latent information, we propose HPPNet that using the Harmonic Dilated Convolution to capture the harmonic structures and the Frequency Grouped Recurrent Neural Network to model the pitch-invariance over time. Experimental results on the MAESTRO dataset show that our piano transcription system achieves state-of-the-art performance both in frame and note scores (frame F1 93.15%, note F1 97.18%). Moreover, the model size is much smaller than the previous state-of-the-art deep learning models.
@misc{wei_hppnet_2022,
title = {{HPPNet}: {Modeling} the {Harmonic} {Structure} and {Pitch} {Invariance} in {Piano} {Transcription}},
shorttitle = {{HPPNet}},
url = {http://arxiv.org/abs/2208.14339},
doi = {10.48550/arXiv.2208.14339},
abstract = {While neural network models are making significant progress in piano transcription, they are becoming more resource-consuming due to requiring larger model size and more computing power. In this paper, we attempt to apply more prior about piano to reduce model size and improve the transcription performance. The sound of a piano note contains various overtones, and the pitch of a key does not change over time. To make full use of such latent information, we propose HPPNet that using the Harmonic Dilated Convolution to capture the harmonic structures and the Frequency Grouped Recurrent Neural Network to model the pitch-invariance over time. Experimental results on the MAESTRO dataset show that our piano transcription system achieves state-of-the-art performance both in frame and note scores (frame F1 93.15\%, note F1 97.18\%). Moreover, the model size is much smaller than the previous state-of-the-art deep learning models.},
urldate = {2022-10-13},
publisher = {arXiv},
author = {Wei, Weixing and Li, Peilin and Yu, Yi and Li, Wei},
month = aug,
year = {2022},
keywords = {Best SoA},
}
Downloads: 0
{"_id":"KNmrsMQZwxDsaZLup","bibbaseid":"wei-li-yu-li-hppnetmodelingtheharmonicstructureandpitchinvarianceinpianotranscription-2022","author_short":["Wei, W.","Li, P.","Yu, Y.","Li, W."],"bibdata":{"bibtype":"misc","type":"misc","title":"HPPNet: Modeling the Harmonic Structure and Pitch Invariance in Piano Transcription","shorttitle":"HPPNet","url":"http://arxiv.org/abs/2208.14339","doi":"10.48550/arXiv.2208.14339","abstract":"While neural network models are making significant progress in piano transcription, they are becoming more resource-consuming due to requiring larger model size and more computing power. In this paper, we attempt to apply more prior about piano to reduce model size and improve the transcription performance. The sound of a piano note contains various overtones, and the pitch of a key does not change over time. To make full use of such latent information, we propose HPPNet that using the Harmonic Dilated Convolution to capture the harmonic structures and the Frequency Grouped Recurrent Neural Network to model the pitch-invariance over time. Experimental results on the MAESTRO dataset show that our piano transcription system achieves state-of-the-art performance both in frame and note scores (frame F1 93.15%, note F1 97.18%). Moreover, the model size is much smaller than the previous state-of-the-art deep learning models.","urldate":"2022-10-13","publisher":"arXiv","author":[{"propositions":[],"lastnames":["Wei"],"firstnames":["Weixing"],"suffixes":[]},{"propositions":[],"lastnames":["Li"],"firstnames":["Peilin"],"suffixes":[]},{"propositions":[],"lastnames":["Yu"],"firstnames":["Yi"],"suffixes":[]},{"propositions":[],"lastnames":["Li"],"firstnames":["Wei"],"suffixes":[]}],"month":"August","year":"2022","keywords":"Best SoA","bibtex":"@misc{wei_hppnet_2022,\n\ttitle = {{HPPNet}: {Modeling} the {Harmonic} {Structure} and {Pitch} {Invariance} in {Piano} {Transcription}},\n\tshorttitle = {{HPPNet}},\n\turl = {http://arxiv.org/abs/2208.14339},\n\tdoi = {10.48550/arXiv.2208.14339},\n\tabstract = {While neural network models are making significant progress in piano transcription, they are becoming more resource-consuming due to requiring larger model size and more computing power. In this paper, we attempt to apply more prior about piano to reduce model size and improve the transcription performance. The sound of a piano note contains various overtones, and the pitch of a key does not change over time. To make full use of such latent information, we propose HPPNet that using the Harmonic Dilated Convolution to capture the harmonic structures and the Frequency Grouped Recurrent Neural Network to model the pitch-invariance over time. Experimental results on the MAESTRO dataset show that our piano transcription system achieves state-of-the-art performance both in frame and note scores (frame F1 93.15\\%, note F1 97.18\\%). Moreover, the model size is much smaller than the previous state-of-the-art deep learning models.},\n\turldate = {2022-10-13},\n\tpublisher = {arXiv},\n\tauthor = {Wei, Weixing and Li, Peilin and Yu, Yi and Li, Wei},\n\tmonth = aug,\n\tyear = {2022},\n\tkeywords = {Best SoA},\n}\n\n\n\n","author_short":["Wei, W.","Li, P.","Yu, Y.","Li, W."],"key":"wei_hppnet_2022","id":"wei_hppnet_2022","bibbaseid":"wei-li-yu-li-hppnetmodelingtheharmonicstructureandpitchinvarianceinpianotranscription-2022","role":"author","urls":{"Paper":"http://arxiv.org/abs/2208.14339"},"keyword":["Best SoA"],"metadata":{"authorlinks":{}},"html":""},"bibtype":"misc","biburl":"https://bibbase.org/zotero/fsimonetta","dataSources":["pzyFFGWvxG2bs63zP"],"keywords":["best soa"],"search_terms":["hppnet","modeling","harmonic","structure","pitch","invariance","piano","transcription","wei","li","yu","li"],"title":"HPPNet: Modeling the Harmonic Structure and Pitch Invariance in Piano Transcription","year":2022}