Python-based, browser-accessible voice-model training platform for creating custom AI singing voices using ACE-Step 1.5 XL. Supports multiple independent voice characters and paired audio recordings, using full song mixdowns with instrumental backing for timing and musical alignment while utilizing matching isolated vocal stems for clean, interference-free model training. Features automated audio segmentation, lyric and caption annotation, listening review, versioned training datasets, configurable LoRA fine-tuning, training job management, real-time logging, and model adapter exports for AI-powered music generation and voice synthesis. Includes dataset preparation and export workflows for SVC/RVC/Applio singing voice conversion, enabling voice transformation while preserving original melodies, timing, and vocal performances. Designed for local, privacy-preserving AI voice development, character modeling, and repeatable training experimentation.
Updated 2026-11-17 06:10:47 +00:00
Python-based GraphRAG platform that transforms collected web content and documents into structured, searchable knowledge graphs for AI-powered research and intelligent question answering. Combines entity and relationship extraction, knowledge reconciliation, community detection, PostgreSQL/PgVector semantic search, and graph-aware retrieval to generate context-rich, source-cited answers. Features a durable, restartable 10-stage processing pipeline, FastAPI services, and portable containerized deployment using Docker or Podman Compose, with an architecture adaptable to Kubernetes. Supports local and cloud-hosted OpenAI-compatible LLMs and integrates with external MCP-based data collection workflows.
Updated 2026-10-29 06:56:57 +00:00
Python implementation of an MCP-powered music composition and groove-building platform for Cubase and other DAWs. Enables AI agents and local or cloud-hosted models to generate deterministic guitar riffs, bass lines, and drum arrangements from detailed text instructions, supporting exact notes, chord voicings, rhythmic patterns, triplets, sustained notes, palm muting, pinch harmonics, articulations, and MIDI CC automation. Generates Standard MIDI Type 1 (960 PPQ), native Ample Sound Riffer (.griff/.briff), and Superior Drummer 3 (.sd3p) files. Supports instrument profiles for Ample Metal Eclipse, Metal Ray5, Guitar LP/M/T, Bass J, and Superior Drummer 3. Designed for integration with MCP-compatible AI platforms, including LM Studio, Ollama-based applications, GitHub Copilot, Visual Studio, and VS Code.
Updated 2026-10-29 05:45:52 +00:00