Decided on Parakeet ONNX Runtime. Works pretty great. Realtime voice chat possible now. UX lacking.

docker-compose.yml

@@ -78,20 +78,18 @@ services:
 
   miku-stt:
     build:
-      context: ./stt
-      dockerfile: Dockerfile.stt
+      context: ./stt-parakeet
+      dockerfile: Dockerfile
     container_name: miku-stt
     runtime: nvidia
     environment:
-      - NVIDIA_VISIBLE_DEVICES=0  # GTX 1660 (same as Soprano)
+      - NVIDIA_VISIBLE_DEVICES=0  # GTX 1660
       - CUDA_VISIBLE_DEVICES=0
       - NVIDIA_DRIVER_CAPABILITIES=compute,utility
-      - LD_LIBRARY_PATH=/usr/local/lib/python3.10/dist-packages/nvidia/cudnn/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
     volumes:
-      - ./stt:/app
-      - ./stt/models:/models
+      - ./stt-parakeet/models:/app/models  # Persistent model storage
     ports:
-      - "8001:8000"
+      - "8766:8766"  # WebSocket port
     networks:
       - miku-voice
     deploy:
@@ -102,6 +100,7 @@ services:
               device_ids: ['0']  # GTX 1660
               capabilities: [gpu]
     restart: unless-stopped
+    command: ["python3.11", "-m", "server.ws_server", "--host", "0.0.0.0", "--port", "8766", "--model", "nemo-parakeet-tdt-0.6b-v3"]
 
   anime-face-detector:
     build: ./face-detector