Interactive demos that work with Lemonade Server.

• realtime_transcription.py: Stream microphone audio for real-time transcription (+ test mode for WAV files)

• llm-debate.html: A debate arena where multiple LLMs can debate each other on any topic
• multi-model-tester.html: Test prompts across multiple models side-by-side

• api_image_generation.py: Generate images using the images/generations endpoint

• api_text_to_speech.py: Generate speech from a prompt using audio/speech endpoint

1. Install Lemonade Server from the latest release
2. Start the server: lemonade-server serve
3. Pull a model if needed (e.g., lemonade-server-dev pull Whisper-Tiny)

Uses the OpenAI-compatible WebSocket API for real-time speech-to-text.

# Install dependencies
pip install openai websockets pyaudio

# Stream from microphone
python realtime_transcription.py

# Use a different model
python realtime_transcription.py --model Whisper-Small

Open the HTML files directly in your browser.

See debate-arena.md for detailed instructions on the debate demo.