Initial implementation of AudioCraft Studio

Complete web interface for Meta's AudioCraft AI audio generation:

- Gradio UI with tabs for all 5 model families (MusicGen, AudioGen,
  MAGNeT, MusicGen Style, JASCO)
- REST API with FastAPI, OpenAPI docs, and API key auth
- VRAM management with ComfyUI coexistence support
- SQLite database for project/generation history
- Batch processing queue for async generation
- Docker deployment optimized for RunPod with RTX 4090

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

config/models.yaml

@@ -0,0 +1,151 @@
+# AudioCraft Model Registry Configuration
+# This file defines all available models and their configurations
+
+models:
+  musicgen:
+    enabled: true
+    display_name: "MusicGen"
+    description: "Text-to-music generation with optional melody conditioning"
+    default_variant: medium
+    variants:
+      small:
+        hf_id: facebook/musicgen-small
+        vram_mb: 1500
+        max_duration: 30
+        description: "Fast, lightweight model (300M params)"
+      medium:
+        hf_id: facebook/musicgen-medium
+        vram_mb: 5000
+        max_duration: 30
+        description: "Balanced quality and speed (1.5B params)"
+      large:
+        hf_id: facebook/musicgen-large
+        vram_mb: 10000
+        max_duration: 30
+        description: "Highest quality, slower (3.3B params)"
+      melody:
+        hf_id: facebook/musicgen-melody
+        vram_mb: 5000
+        max_duration: 30
+        conditioning:
+          - melody
+        description: "Melody-conditioned generation (1.5B params)"
+      stereo-small:
+        hf_id: facebook/musicgen-stereo-small
+        vram_mb: 1800
+        max_duration: 30
+        channels: 2
+        description: "Stereo output, fast (300M params)"
+      stereo-medium:
+        hf_id: facebook/musicgen-stereo-medium
+        vram_mb: 6000
+        max_duration: 30
+        channels: 2
+        description: "Stereo output, balanced (1.5B params)"
+      stereo-large:
+        hf_id: facebook/musicgen-stereo-large
+        vram_mb: 12000
+        max_duration: 30
+        channels: 2
+        description: "Stereo output, highest quality (3.3B params)"
+      stereo-melody:
+        hf_id: facebook/musicgen-stereo-melody
+        vram_mb: 6000
+        max_duration: 30
+        channels: 2
+        conditioning:
+          - melody
+        description: "Stereo melody-conditioned (1.5B params)"
+
+  audiogen:
+    enabled: true
+    display_name: "AudioGen"
+    description: "Text-to-sound effects generation"
+    default_variant: medium
+    variants:
+      medium:
+        hf_id: facebook/audiogen-medium
+        vram_mb: 5000
+        max_duration: 10
+        description: "Sound effects generator (1.5B params)"
+
+  magnet:
+    enabled: true
+    display_name: "MAGNeT"
+    description: "Fast non-autoregressive music generation"
+    default_variant: medium-10secs
+    variants:
+      small-10secs:
+        hf_id: facebook/magnet-small-10secs
+        vram_mb: 1500
+        max_duration: 10
+        description: "Fast 10-second clips (300M params)"
+      medium-10secs:
+        hf_id: facebook/magnet-medium-10secs
+        vram_mb: 5000
+        max_duration: 10
+        description: "Quality 10-second clips (1.5B params)"
+      small-30secs:
+        hf_id: facebook/magnet-small-30secs
+        vram_mb: 1800
+        max_duration: 30
+        description: "Fast 30-second clips (300M params)"
+      medium-30secs:
+        hf_id: facebook/magnet-medium-30secs
+        vram_mb: 6000
+        max_duration: 30
+        description: "Quality 30-second clips (1.5B params)"
+
+  musicgen-style:
+    enabled: true
+    display_name: "MusicGen Style"
+    description: "Style-conditioned music generation from reference audio"
+    default_variant: medium
+    variants:
+      medium:
+        hf_id: facebook/musicgen-style
+        vram_mb: 5000
+        max_duration: 30
+        conditioning:
+          - style
+        description: "Style transfer from reference audio (1.5B params)"
+
+  jasco:
+    enabled: true
+    display_name: "JASCO"
+    description: "Chord and drum-conditioned music generation"
+    default_variant: chords-drums-400M
+    variants:
+      chords-drums-400M:
+        hf_id: facebook/jasco-chords-drums-400M
+        vram_mb: 2000
+        max_duration: 10
+        conditioning:
+          - chords
+          - drums
+        description: "Chord/drum control, fast (400M params)"
+      chords-drums-1B:
+        hf_id: facebook/jasco-chords-drums-1B
+        vram_mb: 4000
+        max_duration: 10
+        conditioning:
+          - chords
+          - drums
+        description: "Chord/drum control, higher quality (1B params)"
+
+# Default generation parameters
+defaults:
+  generation:
+    duration: 10
+    temperature: 1.0
+    top_k: 250
+    top_p: 0.0
+    cfg_coef: 3.0
+
+# VRAM thresholds for warnings
+vram:
+  warning_threshold: 0.85  # 85% utilization warning
+  critical_threshold: 0.95  # 95% utilization critical
+
+# Presets are loaded from data/presets/*.yaml
+presets_dir: "./data/presets"