Realtime voice support demo built on Deepgram Voice Agent + Aura sandbox TTS.
- Vapi-style customer support flow
- Sandbox voice only (
aura-3-canopy-en) - Optional progressive TTS (
/speak+/add_text) - Expressive mode toggle in UI
- Debug panel with latency and flow events
- Create
.envfrom.env.example - Set
DEEPGRAM_API_KEY - Install deps and run
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python client.pyOpen http://127.0.0.1:5051.
- Login:
flyctl auth login- Pick a unique app name and create app:
flyctl apps create <your-unique-app-name>- Update
fly.toml:
- Set
app = '<your-unique-app-name>'
- Set required secret:
flyctl secrets set DEEPGRAM_API_KEY=<your_key>- Deploy:
flyctl deploy- Open:
flyctl openYour public URL will be:
https://<your-unique-app-name>.fly.dev