L'IA peut-elle composer une symphonie dans le style de Mozart indistinguable d'une œuvre perdue authentique ?

L'IA a déjà généré de la musique qui imite de près des compositeurs célèbres. Les modèles récents produisent désormais des œuvres orchestrales à plusieurs mouvements avec une cohérence structurelle. Ils peuvent émuler l'instrumentation baroque, les progressions harmoniques et même les pratiques de performance historiques. Ces œuvres sont souvent débattues par les musicologues quant à leur authenticité.

Background

Recent AI models have demonstrated the ability to generate multi-movement orchestral works with structural coherence, emulating Baroque instrumentation, harmonic progressions, and historical performance practices. Some AI-generated works have been composed with assistance that closely resemble classical-era surface features, such as phrasing, cadences, and orchestration patterns. These advancements have sparked intense debate among musicologists regarding their authenticity and the potential for deception in expert circles.

However, producing a full symphony indistinguishable from an authentic lost Mozart work remains beyond current technological reach. Critics argue that existing systems lack a deep, internalized understanding of Mozart’s compositional process, structural subtleties, and historical context—elements critical to achieving true authenticity. While AI can replicate stylistic conventions, it has yet to demonstrate the originality, emotional depth, and cultural resonance integral to Mozart’s lost compositions. Expert analysis would likely uncover detectable anomalies or anachronisms, betraying the work’s artificial origins.