Talking AI is a simple Node.js application that allows you to upload an MP3 file, convert the speech to text using OpenAI's Whisper API, generate an intelligent answer using OpenAI GPT, and finally convert the generated answer back into speech for playback. This app is designed with a basic front-end and demonstrates a clear chain of AI-based interactions,starting from voice, moving through natural language understanding, and returning to voice.
The aim of this project is to demonstrate how voice-driven AI pipelines can be built using modern tools like Whisper and GPT. It helps developers explore how human-like interaction can be achieved with audio input and output. This can be extended for use cases such as voice assistants, learning tools, customer support bots, and more,especially for Turkish or multilingual audiences.
- Upload MP3 audio files
- Transcribe speech to Turkish text using OpenAI Whisper
- Generate a response based on the transcribed text using OpenAI GPT
- Convert the response back to voice using OpenAI Text-to-Speech
- Simple UI to trigger each step manually
-
Clone the repository:
git clone https://github.com/rastmob/talking-ai.git cd talking-ai
-
Install dependencies:
npm install
-
Create a
.env
file in the root directory and add your OpenAI API key:OPENAI_API_KEY=your-api-key-here
-
Start the server:
npm start
-
Open your browser and go to
http://localhost:3000
- Node.js + Express
- OpenAI Whisper API (Speech to Text)
- OpenAI GPT API (Text Generation)
- OpenAI TTS API (Text to Speech)
- HTML, CSS, JavaScript (Client Side)
talking-ai/
├── public/ # Frontend HTML & JS
├── responses/ # Generated audio files
├── uploads/ # Uploaded MP3s
├── server.js # Express backend
├── .env # Environment variables (not committed)
├── package.json
This project is developed and maintained by Rast Mobile, an innovative software company that specializes in mobile development, AI integrations, and custom web platforms.
Contact & Profiles:
- Website: https://rastmobile.com
- LinkedIn: linkedin.com/company/rastmobile
- GitHub: @rastmob
- Author: @mobilerast, mehmet.alp@rastmobile.com
This project is licensed under the MIT License.