Teach Your AI Composer to Feel Tension
What if you could sketch a “tension curve” and have an AI compose music that follows it? That’s exactly what this study delivers.
Tonal tension—the ebb and flow of stability vs. suspense in harmony—drives how music feels. The authors plug a computational tension model into a Transformer and steer generation with a dual-level beam search:
- Token level: re-ranks notes by model probability and diversity to keep quality high.
- Bar level: re-ranks phrases to match your target tension over time.
In tests, the system reliably modulates tension to fit a desired curve, and listeners confirm the results. Bonus: it can produce multiple, distinct interpretations under the same tension setting—useful for film scoring, games, education, and creative exploration.
Paper by Maral Ebrahimzadeh, Gilberto Bernardes, and Sebastian Stober. Read more: https://arxiv.org/abs/2511.19342v1
Paper: https://arxiv.org/abs/2511.19342v1
Register: https://www.AiFeta.com
AIMusic MusicGeneration GenerativeAI Transformer MusicTech ML DeepLearning MIR Creativity Soundtrack