MAGDA Core: Open-Source DAW with Integrated AI Features

What changed

Conceptual Machines has introduced MAGDA Core, a free and open-source Digital Audio Workstation (DAW) that integrates AI functionalities from its foundation. Developed using C++20, the JUCE framework, and the Tracktion Engine, MAGDA Core aims to facilitate automation, transformation, and rapid musical iteration. The latest release, v0.9.0, brings several key features to the platform. These include hybrid tracks capable of hosting both audio and MIDI clips, three distinct views (Arrangement, Session, and Mix), and a novel AI chat feature. This AI chat allows users to generate and execute custom Domain-Specific Languages (DSLs) directly within the application using natural language commands, supporting a Bring Your Own API Keys (BYOAK) model. The DAW also boasts a robust modulation system with 16 LFOs and 16 macro knobs per device and rack, alongside fully nestable parallel processing racks. Other features include a piano roll with pitchbend and MIDI CC lanes, a drum grid device, a session view for clip launching, and a comprehensive mixer with extensive routing capabilities. The user interface is designed to be collapsible and resizable, with dedicated panels for Inspector, AI Chat, Plugin Browser, Sample Browser, and a context-sensitive editor.

Building MAGDA Core requires a C++20 compiler (GCC 10+, Clang 12+, or Xcode), CMake 3.20+, and Git LFS for handling binary assets. Specific build instructions are provided for Windows, macOS, and Linux, including setup for vcpkg on Windows to manage dependencies like libxml2. The project includes automated GitHub Actions workflows for continuous integration, security scanning (CodeQL, secret detection, vulnerability scanning), and periodic code analysis to identify technical debt. Security is a priority, with branch protection, automated scanning, dependency monitoring via Dependabot, and CI/CD security checks.

The architecture of MAGDA Core is divided into daw/ for the main application logic and agents/ for its AI agent system, with supporting directories for tests, scripts, and documentation. Key dependencies include Tracktion Engine for audio processing, JUCE for the application framework, juce-llm for LLM API interaction, llama.cpp for local LLM inference, and Catch2 for testing.

Why it matters for builders

MAGDA Core's open-source nature and its focus on AI integration present a significant opportunity for AI builders. Developers can explore and contribute to a C++ based audio production environment that natively supports AI-driven workflows. The inclusion of an AI chat interface for DSL generation and the underlying architecture supporting local LLM inference via llama.cpp are particularly relevant for those working on agent systems or natural language interfaces for creative tools. The project's commitment to security and automated workflows also provides a model for robust development practices.

Practical impact

AI builders interested in audio and music production can clone the MAGDA Core repository and explore its codebase. The project provides clear build instructions and make targets for setup, debugging, and release builds. Experimentation with the AI chat feature, understanding how natural language commands translate into DSL, and potentially integrating custom LLMs or agent logic are actionable next steps. Developers can also contribute to the project by filing bug reports or feature requests through GitHub issues, helping to shape the future of this AI-enhanced DAW.

Caveats and source limits

MAGDA Core is described as being in early v0 development, with development having started in January 2026. The project acknowledges that bugs and missing features are expected. While the source mentions AI signals and developer signals, specific details on the performance or capabilities of these AI features are not provided. The license is GPL v3, which builders should consider for any derivative works. No independent benchmarks or performance metrics for the AI components are available in the provided source.