Skip to content

README.md

Latest commit

 

History

History
70 lines (56 loc) · 3.16 KB

README.md

File metadata and controls

70 lines (56 loc) · 3.16 KB
Raw

Songline Feature

Purpose: Living definitions through conversation audio clips - capture organic explanations of DreamNodes from real conversations.

Core Concept

Instead of static text definitions, Songlines preserve authentic voice and context by extracting audio clips from conversations that define/describe DreamNodes. Each "perspective" is a sovereign audio clip stored directly in the DreamNode's repository.

Directory Structure

songline/
├── services/                           # Service layer
│   ├── perspective-service.ts          # Manages perspectives.json files
│   ├── conversations-service.ts        # Manages conversation history
│   ├── audio-recording-service.ts      # Recording path coordination
│   └── audio-trimming-service.ts       # ffmpeg integration for clip extraction
├── components/                         # React components
│   ├── PerspectivesSection.tsx         # Perspectives collection UI
│   ├── AudioClipPlayer.tsx             # Individual clip playback
│   └── ConversationsSection.tsx        # Full conversation history UI
├── index.ts                            # Barrel export
└── README.md                           # This file

Main Exports

Services

  • PerspectiveService - Read/write perspectives, timestamp conversion utilities
  • ConversationsService - Load conversations and transcripts with caching
  • AudioRecordingService - Path generation and directory management for recordings
  • AudioTrimmingService - ffmpeg wrapper for temporal slicing

Components

  • PerspectivesSection - Displays collection of perspective audio clips
  • AudioClipPlayer - Playback UI with transcript toggle
  • ConversationsSection - Full conversation history with lazy loading

Data Structure

Dream Nodes (Ideas)

DreamNode/
├── .udd                    # UUID, title, type, dreamTalk, relationships
├── perspectives.json       # Perspectives from conversations about this idea
└── perspectives/           # Trimmed audio clips (sovereign storage)
    └── {peer}~{me}-{timestamp}.mp3

Dreamer Nodes (People)

DreamerNode/
├── .udd                    # UUID, name, type, contact info
└── conversations/          # Full conversation recordings
    ├── conversation-{date}-{time}.mp3   # Full audio recording
    └── transcript-{date}-{time}.md      # Whisper transcription

Storage Philosophy

  • Full recordings stored in the DreamerNode (who you talked to)
  • Trimmed clips stored in the DreamNode (what you talked about)
  • Each perspective clip is sovereign - physically trimmed, not temporally masked

Notes

  • Sovereign Storage: Each perspective clip is physically trimmed and stored directly in the relevant DreamNode's repository
  • Temporal Slicing: Uses ffmpeg to create independent audio files (not temporal masking)
  • Lazy Loading: Perspectives only load when node is selected in liminal-web layout
  • Caching: Services implement in-memory caching to avoid redundant file I/O
  • Singleton Pattern: All services use singleton instances initialized at plugin startup