​
LLM Model

This example takes two inputs, emotional state and voice inputs, sends them to an LLM, and produces speech outputs and physical movements. The overall behavior of the system is configured in /config/cubly.json5

​
Emotion Detection with Cubly

​
Command

LLM Models
uv run src/run.py cubly

​
Response

Here is typical debug log from the running system:

WARNING:root:Unitree SDK not found. Please install the Unitree SDK to use this plugin.
WARNING:root:Unitree SDK not found. Please install the Unitree SDK to use this plugin.
2025-02-22 09:11:07.365 Python[4339:86735] WARNING: AVCaptureDeviceTypeExternal is deprecated for Continuity Cameras. Please use AVCaptureDeviceTypeContinuityCamera and add NSCameraUseContinuityCameraDeviceType to your Info.plist.
INFO:root:Found cam(0)
INFO:om1_speech:Found 7 audio devices
INFO:om1_speech:Default input device: MacBook Pro Microphone (1)
INFO:om1_speech:Selected input device: MacBook Pro Microphone (1)
INFO:om1_utils:WebSocket client thread started
INFO:om1_speech:Started audio processing thread
INFO:om1_speech:Started audio stream with device 1
INFO:om1_utils:Registered message callback
INFO:root:Initializing OpenAI client with {'base_url': 'https://api.openmind.org/api/core/openai', 'api_key': 'openmind_free'}
INFO:root:Initializing WebSim...
INFO:root:Starting WebSim server thread...
INFO:om1_utils:Connected to wss://api-asr.openmind.org
INFO:om1_utils:Connection established
INFO:root:WebSim server started successfully - Open http://localhost:8000 in your browser
INFO:om1_speech:Found 7 audio devices
INFO:om1_speech:Device 0: LG FHD
INFO:om1_speech:Max Output Channels: 2
INFO:om1_speech:Selected output device: LG FHD at index 0
INFO:om1_speech:Successfully opened audio stream
INFO:root:EmotionCapture: I do not see anyone, so I can't estimate their emotion.
INFO:root:Detected ASR message: yeah
INFO:root:Detected ASR message: so you fix
INFO:root:Detected ASR message: fix the light mode
INFO:root:EmotionCapture: I see a person. Their emotion is neutral.
INFO:root:EmotionCapture: I see a person. Their emotion is happy.
INFO:httpx:HTTP Request: POST https://api.openmind.org/api/core/openai/chat/completions "HTTP/1.1 401 Unauthorized"
ERROR:root:Error asking LLM: Error code: 401 - {'error': 'API key not found'}
WARNING:root:No output from LLM
INFO:root:EmotionCapture: I see a person. Their emotion is neutral.
INFO:httpx:HTTP Request: POST https://api.openmind.org/api/core/openai/chat/completions "HTTP/1.1 401 Unauthorized"
INFO:root:EmotionCapture: I see a person. Their emotion is neutral.
ERROR:root:Error asking LLM: Error code: 401 - {'error': 'API key not found'}
WARNING:root:No output from LLM
INFO:root:Detected ASR message: gussie motion is neutral
INFO:root:EmotionCapture: I see a person. Their emotion is neutral.
INFO:httpx:HTTP Request: POST https://api.openmind.org/api/core/openai/chat/completions "HTTP/1.1 401 Unauthorized"
ERROR:root:Error asking LLM: Error code: 401 - {'error': 'API key not found'}
WARNING:root:No output from LLM
INFO:root:EmotionCapture: I see a person. Their emotion is neutral.
INFO:root:EmotionCapture: I see a person. Their emotion is neutral.
INFO:httpx:HTTP Request: POST https://api.openmind.org/api/core/openai/chat/completions "HTTP/1.1 401 Unauthorized"
ERROR:root:Error asking LLM: Error code: 401 - {'error': 'API key not found'}
WARNING:root:No output from LLM
INFO:root:EmotionCapture: I see a person. Their emotion is neutral.
INFO:root:Detected ASR message: piko found

You should see your webcam light turn on and Cubly should speak to you from your default laptop speaker. You can see what is happening in the WebSim simulator window.

  • There will be an initial delay for your system to download various packages and AI/ML models.

  • Arduino based movement generation only works if you actually have a suitable actuator connected to an Arduino, which is to your computer via a USB serial dongle. On Mac, you can determine the correct serial port name to use via ls /dev/cu.usb*. If you do not specify your computer’s serial port, the example will provide logging data that simulates what it would send.

  • The emotion estimation input is provided by webcam data feeding into cv2.CascadeClassifier(haarcascade_frontalface_default). See /inputs/plugins/webcam_to_face_emotion.

  • The voice input originates from the default microphone, which sends to data to cloud instance of Nvidia’s RIVA. See /inputs/plugins/asr.

  • The voice output uses a cloud text to speech endpoint, which sends audio data to the default speaker. See /actions/speak/connector/tts.

  • The Arduino serial movement actions are sent to serial com port COM1, flowing to a connected Arduino, which can then generate servo commands. See /actions/move_serial_arduino/connector/serial_arduino.