Sprache - Learn German with AI

Sprache is a web application designed to help users learn German using the power of AI. The application leverages OpenAI's text-to-speech capabilities to provide an interactive learning experience.
Its basically built on top of OpenAI's Whisper API.

Table of Contents

• Installation
• Usage
• API
  • POST /api/v1/speech
• File Structure

Installation

1. Clone the repository:

   git clone [email protected]:shadmeoli/sprache.ai.git
   cd sprache

2. Install dependencies:

   bun install

3. Set up environment variables: Create a .env file in the root of the project and add your OpenAI API key:

   GEMMINIAPI_KEY=your-openai-api-key

4. Run the development server:

   bun run dev

   Open http://localhost:3000 with your browser to see the result.

Usage

Navigate to the homepage, where you can interact with the AI-powered German learning tool. Input your text prompts and receive audio responses in German.

API

POST /api/v1/speech

This endpoint generates a speech audio file from a given text prompt using OpenAI's text-to-speech model.

Request

• URL: /api/v1/speech
• Method: POST
• Headers: Content-Type: application/json
• Body:

  {
    "prompt": "Your text prompt here"
  }

Response

• Success:
  • Status: 200 OK
  • Body: Audio file in mp3 format.
  • Headers: Content-Type: audio/mpeg
• Error: Error: 429 Too Many Requests, if the API rate limit is exceeded. This basically happends if you've not payed for the OpenAI API.
  • Body:

    {
      "message": "🥺 Looks like I've not paid, Ooops!"
    }

Example

Request:

curl -X POST http://localhost:3000/api/v1/speech -H "Content-Type: application/json" -d '{"prompt": "Hallo, wie geht es Ihnen?"}'

Response: An audio file in mp3 format is returned.

File Structure

The project directory is structured as follows:

├── package.json
├── postcss.config.cjs
├── prettier.config.js
├── prisma
│   ├── db.sqlite
│   └── schema.prisma
├── public
│   ├── book.png
│   └── favicon.ico
├── README.md
├── src
│   ├── app
│   │   ├── api
│   │   │   └── v1
│   │   │       └── speech
│   │   │           └── route.tsx
│   │   ├── layout.tsx
│   │   └── page.tsx
│   ├── assets
│   ├── audio
│   ├── components
│   │   └── ui
│   │       ├── input.tsx
│   │       ├── sonner.tsx
│   │       └── textarea.tsx
│   ├── env.js
│   ├── lib
│   │   └── utils.ts
│   ├── server
│   │   └── db.ts
│   └── styles
│       └── globals.css
├── tailwind.config.ts
└── tsconfig.json

Key Files and Directories

• src/app/api/v1/speech/route.tsx: API route handler for the speech endpoint.
• src/components/ui: UI components such as input, textarea, and sonner.
• src/styles/globals.css: Global CSS styles.
• prisma/db.sqlite: SQLite database file.
• public: Public assets including images and favicon.

---

I have alse setup a model for logging the prompts and the responses. But I have not implement it yet. Feel if you want to do it, just create a prims client on the api/v1/speech and setup the write.

Don't forget to push your database to create a Local database with sqlite and run the migration.