GACELA: A Generative Adversarial Context Encoder for Long Audio Inpainting of Music

• Published in: IEEE journal of selected topics in signal processing Vol. 15; no. 1; pp. 120 - 131
• Main Authors: Marafioti, Andres, Majdak, Piotr, Holighaus, Nicki, Perraudin, Nathanael
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.01.2021; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Acoustic signal processing; Audio data; Coders; computer generated music; Context; Copying; Discriminators; machine learning; Multiple signal classification; Music; Neural networks; Spectrogram; Synthesizers; Task analysis; Time-domain analysis

In this article, we introduce GACELA, a conditional generative adversarial network (cGAN) designed to restore missing audio data with durations ranging between hundreds of milliseconds and a few seconds, i.e., to perform long-gap audio inpainting. While previous work either addressed shorter gaps or...