OpenAI Workers AI

OpenAI Workers AI is an OpenAI-compatible API hosted on Cloudflare Workers. It leverages Cloudflare Workers AI to process and transcribe audio using various Whisper-based models. The project is built with the Hono framework and provides OpenAPI documentation with Swagger UI.

---

Features

• Cloudflare Workers Integration: Leverages Cloudflare's serverless platform for fast and scalable API hosting.
• OpenAI Compatible API: Supports endpoints that mimic the OpenAI API for AI functionalities.
• Audio Transcription: Implements transcription endpoints that process audio using multiple Whisper models:
  • @cf/openai/whisper
  • @cf/openai/whisper-tiny-en
  • @cf/openai/whisper-large-v3-turbo
• API Documentation: Automatically generated OpenAPI documentation available at /openapi.json and rendered via Swagger UI at /docs.
• Type Safety & Validation: Built with TypeScript and uses Zod for schema validation.

---

Installation

Clone the repository and install the dependencies:

npm install

---

Local Development

To start a local development server using Wrangler:

npm run dev

---

Deployment

Deploy your project to Cloudflare Workers with:

npm run deploy

You can also deploy directly using the Deploy to Cloudflare Workers button at the top.

Note: Replace YOUR_ACCOUNT_ID in the deploy button URL with your actual Cloudflare account ID.

---

API Endpoints

Root Endpoint

• GET /
  Returns a simple greeting message.

Documentation

• GET /docs
  Renders the Swagger UI documentation for the API.

Audio Transcription

• POST /v1/audio/transcriptions
  Accepts a multipart/form-data request with the following fields:
  • file: Audio file to transcribe.
  • model: Select one of the supported models:
    • @cf/openai/whisper
    • @cf/openai/whisper-tiny-en
    • @cf/openai/whisper-large-v3-turbo
  • language (optional): Language of the audio.
  • prompt (optional): Initial prompt to guide the transcription.
  • response_format (optional): Format of the response. Options include json (default), text, srt, verbose_json, and vtt.
  • temperature (optional): Temperature value between 0 and 1 (default is 0).
  • timestamp_granularities[] (optional): Array indicating the granularity of timestamps, either "word" or "segment" (default is ["segment"]).

Note: Currently, only the JSON response format is implemented. Other response formats will return a "Not implemented" error.

---

Project Structure

lqm1-openai-workers-ai/
├── README.md
├── biome.json
├── package.json
├── tsconfig.json
├── worker-configuration.d.ts
├── wrangler.jsonc
└── src/
    ├── index.ts
    └── v1/
        └── audio.ts

• src/index.ts
  Sets up the Hono application, including logging, API documentation, and route mounting.

• src/v1/audio.ts
  Defines the transcription endpoint and handles audio processing based on the selected Whisper model.

---

Contributing

Contributions are welcome! Please open an issue or submit a pull request for improvements or bug fixes.

---

License

This project is licensed under the MIT License.