Musical Model Training
Doug discusses the process of training a model on a diverse collection of classical music pieces, emphasizing the significance of autoregression in composing new scores. He reveals that the model is trained on around 20,000 to 30,000 performances, primarily of classical works, including pieces by Chopin. Additionally, he highlights the potential for musicians to train models on their own playing styles, enabling them to create music that reflects their unique improvisational flair.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Expressive AI-Generated Music With Google's Performance RNN with Doug Eck - #32
Related Questions