TARS-inspired UI with:
- Azure OpenAI for chat (Assistants-style minimal wrapper)
- Azure Speech for voice in/out (push-to-talk mic capture + TTS to speakers)
- Face recognition provider (default: Azure Face; optional: Incoresoft VEZHA® REST adapter)
Fan-made UI inspired by Interstellar's TARS (not affiliated).
python3 -m venv .venv && source .venv/bin/activate
pip install -e .# Azure OpenAI (required)
AZURE_OPENAI_ENDPOINT=https://your-openai-resource.openai.azure.com
AZURE_OPENAI_API_KEY=your-openai-key
AZURE_OPENAI_DEPLOYMENT=gpt-4o-mini # your deployment name
AZURE_OPENAI_API_VERSION=2024-08-01-preview
# Azure Speech (for mic + TTS) (required for voice features)
AZURE_SPEECH_KEY=your-speech-key
AZURE_SPEECH_REGION=your-region
# Face (default = Azure Face; set keys if used)
AZURE_FACE_ENDPOINT=https://<region>.api.cognitive.microsoft.com
AZURE_FACE_KEY=your-azure-face-keytars-ui- Push-to-talk: hold the mic button to capture a single utterance; release to stop and send to chat.
- TTS voice: uses your Azure Speech default voice; change with
AZURE_SPEECH_VOICE(e.g.,en-US-JennyNeural).
- Grant Terminal/Python access to Microphone and Camera in System Settings → Privacy & Security.
- Close other apps that might use the camera/mic.