Majik's Songwriter
Your conversations have stories worth singing.
Songwriter v1.4.0 release.
Most AI songwriting tools generate generic lyrics from a prompt. Majik's Songwriter starts from your actual life — your text messages become raw material, a knowledge graph maps the people, themes, and emotions across your conversations, and 14 songwriting strategies explore your data from different angles.
Everything runs locally on your machine — your conversations never leave your device. Or connect to cloud LLMs via OpenRouter, Ollama, or any OpenAI-compatible API.
Complementary to Music Studio. Songs flow from Songwriter into Music Studio for production, then to majiks.online for distribution. Art Studio generates cover artwork automatically from song metadata.
Chat with Cobrohl
Your AI songwriting collaborator. Describe a mood, mention a friend, or paste a memory. Cobrohl writes lyrics from what it finds in your conversations.
Three Generation Modes
Surprise Me (AI picks the best moments), Guided (pick a person, time, genre, mood), or Full Control (set BPM, key, theme, every parameter). 14 songwriting strategies explore your data from different creative angles.
Refinement Loop
Regenerate any section — chorus, verse, bridge — while keeping the rest. Swap genres, adjust mood, or try a different songwriting strategy without starting over.
Your Conversations as Source
Import iMessage conversations. A knowledge graph maps people, themes, and emotions. Your real life becomes raw material for authentic lyrics.
Knowledge Graph
Maps people, places, emotions, and recurring themes across all your conversations. The graph grows as you add more data — deeper relationships mean richer songs.
Articles & RSS Feeds
Add RSS feeds or paste URLs. Write songs inspired by current events, essays, or any text. Pre-configured feeds from AP, Reuters, BBC, NPR, and more.
Music Theory Built In
AI-generated chord progressions, guitar tabs, MIDI export, MusicXML. Audio analysis detects BPM, key, and chords from imported files.
Audio Analysis
Import audio files and extract BPM, key, chord progressions, and structure. Use detected attributes as constraints for new songs that match existing tracks.
Batch Generation & Queue
Queue songs with different strategies. Generate 10 mixed songs and come back later. Continuous mode keeps generating until you stop it.
Song Library & Albums
Every song saved with full metadata. Organize into albums, search, filter, favorite, tag. In-app editing of lyrics and metadata.
Ecosystem Integration
Studio Bridge sends songs to Music Studio for production. Art Studio generates cover artwork automatically. Pipeline export carries all metadata downstream.
MCP Server
12 MCP tools for Claude Code integration. Generate songs, manage the library, query the knowledge graph, and automate batch workflows programmatically.
Explore & Parameters
Generation Queue
Knowledge Graph
Your text messages mapped as a living network of people, themes, and emotions. The graph grows as you add conversations — deeper relationships mean richer songs.
Library & Lyrics
Text Message Sources
Music Theory & Tabs
Audio Analysis
System Requirements
| Component | Requirement |
|---|---|
| macOS | 15.0 (Sequoia) or later |
| Processor | Apple Silicon recommended (Intel works with cloud LLMs) |
| Memory | 8 GB minimum, 16 GB for local MLX models |
| Storage | ~2 GB for app + models on first use |
Intel Macs work with cloud LLM providers (OpenRouter, Ollama, LM Studio) but cannot run local MLX inference.