Add OmniVoice TTS module with config, API, profiles and CLI

- Create modules/omnivoice/ with VoiceAPI, VoiceProfiles, CLI
- Add config manager integration with local model support
- Add app/komAI.py entry point
- Add tests/test_omnivoice.py
- Clone OmniVoice to external/ for development
- Add omnivoice config to global.yaml

modules/omnivoice/config.py

@@ -0,0 +1,64 @@
+import src.utils.config_manager as config
+
+config = config.config
+
+MODEL_NAME = "model_name"
+MODEL_PATH = "model_path"
+DEVICE = "device"
+DTYPE = "dtype"
+NUM_STEPS = "num_steps"
+SPEED = "speed"
+PROFILES_DIR = "profiles_dir"
+OUTPUT_DIR = "output_dir"
+ENABLED = "enabled"
+
+DEFAULT_MODEL_NAME = "k2-fsa/OmniVoice"
+DEFAULT_MODEL_PATH = "models/OmniVoice"
+DEFAULT_DEVICE = "cuda:0"
+DEFAULT_DTYPE = "float16"
+DEFAULT_NUM_STEPS = 32
+DEFAULT_SPEED = 1.0
+DEFAULT_PROFILES_DIR = "data/voice_profiles"
+DEFAULT_OUTPUT_DIR = "output/voice"
+DEFAULT_ENABLED = True
+
+
+def register_params():
+    config.register(
+        name=MODEL_NAME,
+        val=DEFAULT_MODEL_NAME,
+        cat="omnivoice",
+        desc="Model name (HuggingFace) or use model_path for local",
+    )
+    config.register(
+        name=MODEL_PATH,
+        val=DEFAULT_MODEL_PATH,
+        cat="omnivoice",
+        desc="Local model path (if set, overrides model_name)",
+    )
+    config.register(
+        name=DEVICE, val=DEFAULT_DEVICE, cat="omnivoice", desc="Device for inference"
+    )
+    config.register(name=DTYPE, val=DEFAULT_DTYPE, cat="omnivoice", desc="Data type")
+    config.register(
+        name=NUM_STEPS, val=DEFAULT_NUM_STEPS, cat="omnivoice", desc="Diffusion steps"
+    )
+    config.register(name=SPEED, val=DEFAULT_SPEED, cat="omnivoice", desc="Speed factor")
+    config.register(
+        name=PROFILES_DIR,
+        val=DEFAULT_PROFILES_DIR,
+        cat="omnivoice",
+        desc="Voice profiles directory",
+    )
+    config.register(
+        name=OUTPUT_DIR,
+        val=DEFAULT_OUTPUT_DIR,
+        cat="omnivoice",
+        desc="Output directory for audio",
+    )
+    config.register(
+        name=ENABLED,
+        val=DEFAULT_ENABLED,
+        cat="omnivoice",
+        desc="Enable omnivoice module",
+    )