Voice Thingy

An open source tool - converse with AI in audio. Powered by OpenAI Realtime, Braintrust & Neon.

Tech stack

• Neon for storing conversation history
• OpenAI Realtime API on Braintrust for audio conversation powered by AI
• Next.js with TypeScript for the app framework
• Tailwind for styling

Cloning & running

1. Clone the repo: git clone https://github.com/neondatabase-labs/voice-thingy
2. Create a .env file and add the following environment variables:

# Grab a connection string from https://console.neon.tech
DATABASE_URL="postgresql://neondb_owner:...@ep-...us-east-1.aws.neon.tech/neondb?sslmode=require"

# Grab API Key from https://www.braintrust.dev/app/settings/api-keys
BRAINTRUST_API_KEY="sk-xlKWNORh5P4zOneLDeYq78VQqed9WBmIKSHddl7WDbHXh107"

3. In Braintrust's Settings, set your OPENAI_API_KEY.
4. Run pnpm install and pnpm schema to install dependencies and set the relevant schema in your Neon database.
5. Run pnpm dev to run the application locally.