Skip to main content
Glama
README.md
# AI Producer Hub

Fleet orchestration MCP server for the audio production stack. Generates songs via SongGeneration LeVo, loads them to VirtualDJ decks, produces multi-track albums, and sets up live streaming.

## Quick Start

```powershell
git clone https://github.com/sandraschi/ai-producer-hub
cd ai-producer-hub
uv sync
uv run -m ai_producer_hub    # MCP server (MIDI tools)
```

Webapp: `cd webapp && npm install && npx vite --port 10707`

## Tools

| Tool | What it does |
|------|-------------|
| `songgen_to_deck(lyrics, genre, tempo, deck_id)` | Generate song via LeVo → load to VDJ deck |
| `ai_dj_set(theme, num_tracks, bpm)` | Generate multi-track DJ set across decks |
| `album_factory(theme, num_tracks)` | Multi-track album, optionally to Plex |
| `live_stream_producer(theme, duration)` | Set up streaming with AI-generated music |
| `ai_mashup(track_a, track_b)` | Mashup plan from VDJ library search |
| `hub_status()` | Check all fleet server health |
| `list_midi_devices()` | Enumerate MIDI hardware |
| `record_midi_performance(device, duration)` | Record MIDI to file |

## Architecture

```
AI Producer Hub (MCP stdio)
  ├──→ songgeneration-mcp (HTTP :10885) — LeVo AI generation
  ├──→ virtualdj-mcp (HTTP :10877) — DJ decks, stems, mixing
  ├──→ reaper-mcp (HTTP :10797) — DAW mastering
  └──→ obs-mcp (HTTP :10819) — Live streaming
```

Sub-servers can run on any reachable host. Configure via `SONGGEN_BASE`, `VDJ_BASE`, `REAPER_BASE`, `OBS_BASE` env vars.

## Requirements

- Python 3.11+
- Running sub-servers (songgeneration-mcp, virtualdj-mcp) for orchestration tools
- `python-rtmidi` + `mido` for MIDI tools

TDQS

B3.1/5.0

Scored across 22 tools

Disambiguation3/5

The MIDI tools (record_midi_performance, send_midi_note, play_midi_file, midi_monitor, list_midi_devices) are clearly distinct, but the AI workflow tools overlap heavily: ai_produce_track, ai_orchestrate_production, and ai_colaborate_workflow all describe autonomous AI orchestration, and live_stream_producer vs ai_stream_production appear to do nearly the same thing. Descriptions help somewhat but boundaries between the AI-orchestration tools are genuinely fuzzy.

Naming Consistency4/5

All names use snake_case, which is consistent, and many follow a verb_noun pattern (send_midi_note, play_midi_file, list_midi_devices). There are deviations (ai_* prefixed tools, noun-style album_factory/karaoke_generator/hub_status) and a typo (ai_colaborate_workflow), but the surface remains readable and predictable.

Tool Count3/5

At 22 tools this sits in the heavy range for a server whose purpose is one production pipeline, and several tools (multiple AI-orchestration and streaming variants) feel redundant rather than each earning its place. It is not extreme, but it leans toward over-provisioned.

Completeness4/5

The domain (AI music production and DJ mixing) is covered broadly: MIDI I/O, track/album generation, remix, mashup, karaoke, streaming, and library integration. Minor gaps exist around explicit track editing/mixing controls, but core lifecycle workflows are represented.

Maintenance

ActivityActive
ResponsivenessNo issues