Creating your own musical pieces is one of the most wonderful ways of enjoying music, but many lack the basic musical skills to do so. Andrew Caosun discusses a neural hidden Markov model (NHMM)—a hybrid of a hidden Markov process and a convolutional neural network algorithm with LSTM. This model takes users’ original musical ideas, automatically modifies the input, and generates musically appropriate melodies as output.
The model is extended to allow users to specify magnitude of revision, duration of music segment to be revised, choice of music genres, popularity of songs, and cocreation of songs in social settings. These extensions enhance user understanding of music knowledge, enrich their experience of self-music learning, and enable social aspects of music making. The model is trained using Columbia’s publicly available Million Songs Dataset.
Andrew also explains how he and his team designed a mobile application with an intuitive, interactive, and graphical user interface that’s suitable for the elderly and young children. Unlike most existing literature focusing on computer music composing itself, their research and application aim to use computers to aid human composition and enrich music education for those without musical training.
Andrew Caosun is a senior at Horace Mann School. He’s been actively involved with Concerts in Motion since middle school, spending Sunday afternoons singing with seniors in nursing homes, and has also participated in seasonal events at the Turtle Bay Music School, raising a music education fund for children from disadvantaged families. The friendships he developed during these events helped him understand just how much music can mean to someone, giving him the idea to combine his love for singing and recent technological advancements to help others compose their own pieces. His research, under the guidance of David Gu, applies a hybrid HMM and convolutional neural network with LSTM to compose music.
©2019, O'Reilly Media, Inc. • (800) 889-8969 or (707) 827-7019 • Monday-Friday 7:30am-5pm PT • All trademarks and registered trademarks appearing on oreilly.com are the property of their respective owners. • email@example.com