[1]
S. Li, D. Choi, and Y. Sung, “MIDILM: A Dual-Path Model for Controllable Text-to-MIDI Generation”, AAAI, vol. 40, no. 28, pp. 23160–23168, Mar. 2026.