Speech Streaming, OpenAI, Elevenlabs streaming, Vite, Fastify, ModelFusion Server

Given a prompt, the server returns both a text and a speech stream response.

• ModelFusion Flow: src/flow/duplexStreamingFlow.ts
• Server: src/server/server.ts
• Client: src/App.tsx

1. Create .env file with the following content:

   OPENAI_API_KEY="YOUR_OPENAI_API_KEY"
   ELEVENLABS_API_KEY="YOUR_ELEVENLABS_API_KEY"
   

2. Increase the max completion tokens in duplexStreamingFlow.ts (by default it generates a very short text)

3. Start the server:

   pnpm run server

4. Start the client:

   pnpm run client

5. Go to the URL from the client run