Interactive AI-assisted compositional platform for last-year music project: exploring generative harmony and adaptive orchestration for contemporary film scoring

Project Abstract

This study presents the development and evaluation of an interactive AI-assisted compositional platform designed for final-year music project work, with a focus on exploring generative harmony and adaptive orchestration tailored to contemporary film scoring. The platform integrates deep learning models for harmonic progression and texture synthesis, a rule-based constraint system to ensure stylistic coherence with cinematic narratives, and an adaptive orchestration engine that maps musical decisions to instrument timbres, articulations, and dynamic contours. The research addresses the need for accessible, artist-driven AI tools that augment human creativity while preserving expressive intent and control over musical outcomes. A multi-phase methodology combines literature review, system design, algorithmic development, and iterative user-centered evaluation conducted with a cohort of final-year music students and professional composers. The core technical contribution includes a hierarchical generative model that produces harmonic sequences conditioned on genre-mue cues, narrative mood metadata, and user-provided sketches, plus an orchestration module leveraging a probabilistic instrument-choir mapping to produce orchestrational textures that adapt to scene pacing and emotional arcs. The platform supports interactive manipulation through a visual composer interface, real-time audio rendering, and explainable AI components that articulate rationale for chord choices, voice-leading decisions, and instrumental allocations. The experimental phase assesses musical quality, stylistic alignment with target film genres, computational efficiency, and user satisfaction. Evaluation employs both objective metrics (harmonic consonance/dissonance profiles, voice-leading compactness, orchestration diversity, and temporal coherence relative to scene structure) and subjective measures (expert panel ratings, user-reported ease of use, and perceived creative empowerment). Results indicate that the AI-assisted workflow accelerates idea generation, expands harmonic vocabulary, and yields orchestration textures that maintain cinematic expressivity while reducing cognitive load during ideation. Notably, the adaptive orchestration component demonstrates robust performance in aligning orchestral color, articulation, and dynamics with dynamic scene requirements, even under constrained by-the-scene tempo and mood shifts. The study also investigates ethical and creative implications of AI collaboration in music production, including authorship, originality, and the risk of stylistic homogenization, proposing guidelines for transparent user control, versioning, and provenance tracking. A comparative analysis with baseline non-AI compositional strategies highlights significant improvements in efficiency, novelty, and narrative alignment without compromising technical rigor. The platform's architecture emphasizes modularity and extensibility, enabling future incorporation of additional generative modalities (rhythmic patterns, motif development, and messa di voce dynamics) and domain-specific plug-ins for film scoring pipelines (DAWs, middleware synchronization, and third-party orchestration libraries). The study contributes to the growing discourse on human-AI co-creation in music by offering a tangible, deployable toolset for contemporary scoring contexts, accompanied by an open-access dataset of modeled harmonic progressions, orchestration mappings, and evaluative benchmarks to foster further research and cross-disciplinary collaboration.

Project Overview

What This Project Is About

---

This project looks at how artificial intelligence can assist in creating music and organizing sounds for film scoring. It combines algorithms that can suggest melodies, harmonies, and instrument choices with a workflow that adapts to the mood and pacing of a scene. The goal is to make a flexible tool that helps composers experiment quickly and produce polished ideas without needing to program from scratch.

---

---

The Problem It Addresses

---

Film scoring often requires balancing creative ideas with tight deadlines. Musicians may struggle to generate varied harmonies and instrument textures that fit different scenes. This project aims to reduce the time and effort needed to draft musical ideas while ensuring the music sounds cohesive and professional.

---

---

Objectives of the Project

---

1. Explore how AI can suggest harmonies that fit different emotional tones.
2. Develop a workflow that adapts orchestration to scene dynamics.
3. Create an easy-to-use interface for non-programmers to experiment with ideas.
4. Test the system with sample film scenes and gather feedback from musicians.

---

---

What You Will Do Step by Step

---

1) Review existing music AI tools and identify gaps in harmony and orchestration features. 2) Collect or generate simple musical datasets (short motifs, chord progressions). 3) Build or adapt algorithms to propose harmonies and instrument combinations. 4) Implement a user-friendly interface to run experiments. 5) Evaluate outputs by listening tests and expert feedback. 6) Refine the tool based on results and document the process.

---

---

---

---

Expected Outcome

A functional, user-friendly AI-assisted platform that can propose harmonies and adaptive instrument textures for film cues, along with a short showcase of scored scenes and a user guide. The project should demonstrate faster idea generation and a clearer workflow for composers.