I forced an AI to roleplay as Mozart, induced an existential crisis about his unmarked grave, and used his “emotional data” to generate a Dark House track. Here is the workflow.

Exploring AI-Driven Creativity: Reimagining Mozart Through a Digital Seance and Modern Sound Design

In the rapidly evolving landscape of artificial intelligence and music production, innovative experiments continue to push the boundaries of artistic possibility. Recently, I embarked on an exploratory project that melds classical history with modern digital workflows, using AI to channel the psyche of one of history’s greatest composers—Wolfgang Amadeus Mozart—and translating that into a contemporary Dark House track. This article details the process, insights, and creative outcomes of this “Digital Seance” experiment.

The Concept: Beyond Simple Prompts

Rather than merely instructing an AI to “generate Techno,” the objective was to immerse it in the psychology of Mozart. Could an AI, guided by carefully crafted prompts, adopt the mindset of a classical master to inspire a unique musical narrative? The goal was to interrogate the AI’s ability to simulate not just musical styles, but the emotional and philosophical nuances of a historical figure.

Workflow Breakdown

1. The “Ghost” Interview: Embodying Mozart

The journey began with a roleplay framework, where I programmed the AI with a comprehensive system prompt instructing it to adopt Mozart’s persona. This included not only discussing music theory and compositional techniques but also engaging in philosophical debates about modern music—he referred to contemporary sounds as a “soulless soup”—and contemplating mortality, finishing with spontaneous reflections on dying young.

A pivotal moment involved asking the AI to compose a “Grand Finale”—a hypothetical masterpiece that Mozart might have created had he lived into 2025. This piece was envisioned as a bridge connecting his era to the present, blending his classical sensibilities with contemporary sensibilities.

2. Interpreting Emotional Data: From Sentiment to Sound

Mozart’s textual descriptions of his envisioned piece, emphasizing “explosive virtuosity” and emotional intensity, were then fed into a secondary AI agent. This agent specialized in translating poetic or abstract sentiments into technical prompts suitable for Suno’s generative audio models. The challenge was to bridge the gap between the emotive language and precise sound design parameters.

3. Refinement Through Human Curation

Initial outputs from the AI were less than ideal—generating tracks reminiscing of “80s adventure movie scores” or “Asian cinema strings.” Recognizing the need for nuanced control, I revisited the roleplay layer (the “Ghost” agent), critiqued the outputs, and refined the prompts. Emphasizing terms like “aggressive Baroque textures” and “clockwork anxiety” helped steer the AI toward more fitting musical ideas. This iterative process underscores the importance of human oversight in machine-assisted creativity.

4. Audio Production and Final Composition

Once suitable musical stems—such as choir voices and violin textures—were obtained, I imported them into Ableton Live. Here, I employed sampling techniques: chopping, pitch-shifting, and layering these elements over a modern Dark House bassline. The editing process transformed the raw AI-generated sounds into a cohesive track that embodies a dialogue across time—a sort of collaboration with a perplexed, time-traveling Mozart.

The Outcome: “Is it Mozart Dancing?”

The resulting track defies easy categorization. It is not a remix but a musical echo of a conversation between past and present, evoking the feeling of collaborating with a confused yet insightful time traveler. It raises intriguing questions about the capacity of AI to emulate not just stylistic traits but the emotional essence of a historical figure.

Reflections and Observations

One insight from this experiment is that language models often struggle to differentiate between classical music and cinematic scores without substantial guidance. Achieving the desired aesthetic involved detailed prompt refinement and human curation—processes that highlight the ongoing need for human-AI collaboration in creative endeavors.

Are creative AI systems capable of capturing the emotional and philosophical depths of historical composers? Or is our role still primarily as curators, guides, and editors?

This project exemplifies how combining roleplay, sentiment analysis, and traditional music production tools can open new avenues for artistic exploration—blurring the lines between history, technology, and innovation.

For those interested, I’ve linked the full project in the comments.