Tonefold

Product Introduction

1. Definition: Tonefold is a free, open-source, desktop-based AI music composition and MIDI editing application for macOS and Windows. It functions as a hybrid AI agent and digital audio workstation (DAW) environment, generating structured musical parts from natural language descriptions.
2. Core Value Proposition: Tonefold exists to bridge the gap between AI-assisted music generation and hands-on, professional music production. It enables musicians, producers, and composers to rapidly prototype complete song arrangements with editable MIDI data, offering granular control over every generated note within a traditional piano roll interface.

Main Features

1. AI-Powered Composition with Claude/Ollama: The core feature is a chat-based composer powered by Anthropic's Claude AI (via Claude Code) or a local Ollama model. Users describe a song's mood, style, or specific elements in plain language. The AI agent then generates a structured song with distinct layers (chords, melody, bass, drums, etc.) and sections (intro, verse, chorus), placing all notes into an editable piano roll. This leverages the Claude Agent SDK for tool use and state management.
2. Layer-Based Producer Mode: A specialized "Producer" workflow enforces a structured, approval-based composition process. A producer agent first analyzes the session and proposes a detailed plan (key, tempo, form, and specific layers to modify). The user must approve this plan before four specialist agents (Harmony & Form, Melody & Topline, Rhythm Section, Arrangement & Mix) write only to their assigned layers. This mimics a professional studio workflow and prevents uncontrolled overwrites.
3. Native Desktop Application with Local Processing: Built with Rust for performance, the Tonefold application runs natively, handling the piano roll UI, song state management, playback via a built-in General MIDI synthesizer (GeneralUser GS), and all audio export locally. The AI composer runs as a separate Node.js sidecar process, communicating with the main app via a local WebSocket. This architecture ensures generation, editing, playback, and WAV/MIDI export function entirely offline; only the chat component requires an AI model endpoint.
4. DAW Integration & CLI Tools: Beyond the standalone app, Tonefold offers optional CLAP and VST3 plugin versions for integration into DAWs like FL Studio (the tested host). Additionally, a Command Line Interface (CLI) and a skill file allow AI coding assistants like Claude Code or Cursor to compose music programmatically, enabling automation and integration into larger AI agent workflows.

Problems Solved

1. Pain Point: Many AI music generators act as "black boxes," outputting audio loops or MIDI files with little to no user control over the individual musical components, making integration into a professional production workflow difficult.
2. Target Audience: The primary users are music producers, beatmakers, composers, and songwriters who want to use AI for inspiration and rapid prototyping but require the precision of MIDI editing. It's also valuable for developers and AI enthusiasts interested in agentic workflows for creative tasks.
3. Use Cases: Rapidly sketching a full song arrangement from a text prompt; generating alternative basslines or drum patterns for an existing chord progression; using the "Producer mode" to collaboratively build a track with AI in a structured, section-by-section manner; exporting stems and MIDI to finish a production in a primary DAW; using the CLI to incorporate AI composition into an automated content pipeline.

Unique Advantages

1. Strengths & Limitations (Pros & Cons):

   • Pros:
     • Unprecedented Editability: The core advantage is the generation of fully editable, layer-separated MIDI data, not just audio.
     • Structured, Controllable AI Workflow: Producer mode offers a unique, plan-and-approve system that provides significant user control over the AI's output.
     • Privacy & Offline Capability: Local execution for everything except the model call (which can also be local via Ollama) is a major benefit for data-sensitive and offline workflows.
     • Free and Open Source: Zero cost and full code transparency under GPL-3.0.
     • Flexible Model Support: Works with cloud-based Claude or self-hosted Ollama models.
   • Cons:
     • Technical Barrier: Requires Node.js setup and configuration of Claude Code or an Ollama server for the AI features, which may deter non-technical users.
     • Limited Sound Design: Relies on a basic General MIDI soundfont for playback; users must export MIDI to a full DAW for professional sound design and mixing.
     • New & Unnotarized: As a new project, the macOS build is not notarized by Apple, triggering security warnings on first launch.
     • DAW Plugin Support: Plugin compatibility is currently only confirmed with FL Studio.

2. Key Alternatives & Differentiation:

   • AIVA, Soundraw, Boomy: These are primarily cloud-based AI composition services that generate finished audio tracks or loops. Differentiation: Tonefold is a free, local desktop app focused on generating and editing MIDI data, offering far deeper compositional control and no subscription fees.
   • DAWs with AI Tools (e.g., FL Studio's "Stem Separation," Ableton's "MIDI Devices"): These are built-in features for manipulating existing audio or generating simple patterns. Differentiation: Tonefold is a dedicated, agentic AI composer that generates complete, multi-layered arrangements from text descriptions, functioning as an idea generator rather than a production effect.
   • OpenAI's MuseNet or Google's MusicLM: These are large-scale research models for generating audio. Differentiation: Tonefold is an end-user application with a graphical interface, a structured piano roll editor, and a workflow designed for music production, not just raw audio generation.

Frequently Asked Questions (FAQ)

1. Is Tonefold really free to use? Yes, the Tonefold application is completely free and open-source (GPL-3.0). You own the music you create. Costs may apply if you use the cloud-based Claude AI model through your own subscription; using a local Ollama model is free.
2. Can I use Tonefold without an internet connection? Yes, for generation you can use a local Ollama model. Playback, MIDI editing, and WAV/MIDI export work entirely offline. Only the chat feature requires a connection to an AI model endpoint, which can be on your local network.
3. What audio formats can I export from Tonefold? Tonefold can export the entire project as a Standard MIDI File (.mid) for use in any DAW. It can also render audio to a stereo WAV mix file or individual WAV stem files for each layer (e.g., chords.wav, bass.wav).
4. Do I need a Digital Audio Workstation (DAW) to use Tonefold? No, Tonefold is a fully functional standalone application for composition and editing. The DAW plugins (CLAP/VST3) are optional for users who want to run the Tonefold composer inside their existing production environment.
5. How does Tonefold's "Producer mode" differ from its regular "Composer mode"? Composer mode is for quick, direct AI generation where changes are applied immediately. Producer mode is a structured, multi-agent workflow where a "producer" first proposes a detailed plan for the song's key, form, and which layers to change. The user must approve this plan before specialized AI agents write to the session, providing greater oversight and preventing unwanted changes.