QMUL, School of Electronic Engineering and Computer Science
Centre for Digital Music Seminar Series
Qiuqiang Kong (ByteDance)
Date/time: Wednesday 3rd of November, 10.00-11.00 am
Open to students, staff, alumni, public; all welcome. Admission is FREE, no pre-booking required.
Title: High-resolution piano transcription system andGiantMIDI-Piano dataset
Abstract: The joint research of artificialintelligence (AI) and music has been a popular research area in recent years.AI and music have valuable applications in both academia and industry,including music recommendation, production and automatic generation. Musictranscription is a fundamental task in AI based music tasks, which aims attranscribing audio recordings into symbolic representations. This work proposeda high-resolution piano transcription system including transcribing piano notesand sustain-pedals. All of onset, offsets and velocities information aretranscribed. Using this piano transcription system, we created aGiantMIDI-Piano dataset including 10,854 unique automatically transcribed pianoMIDI files. We analyzed the GiantMIDI-Piano dataset and show musical analysisof some classical music pieces. GiantMIDI-Piano dataset can be used but notlimited to automatic music generation and musical analysis.
Bio: Qiuqiang Kong started his Ph.D. study at Centrefor Digital Music (C4DM), Queen Mary University of London in 2015, and receivedhis Ph.D. degree from University of Surrey, Guildford, UK in 2020. Followinghis PhD, he joined ByteDance AI Lab as a research scientist. His research topicincludes the classification, detection and separation of general sounds andmusic.
Presentation Slides: music_transcription_giantmidi-piano_qiuqiangkong.pdf