ML Interview Platform

Overview

This project is an AI-powered interview platform built with Next.js. It integrates Deepgram for speech-to-text (STT) conversion and Gemini LLM for natural language processing (NLP). The platform facilitates interactive interviews by converting the user's spoken answers into text, generating responses or questions via Gemini LLM, and converting the generated text back into speech for a seamless, voice-driven interview experience.

Project Video: https://youtu.be/NM5Y3exMh4U

Features

• Real-time Voice Input: Users can speak directly into the app, and their voice is converted to text via Deepgram's speech-to-text service.
• Natural Language Processing: The transcribed text is processed by Gemini LLM, which generates meaningful, context-aware responses.
• Voice Response: The response from Gemini LLM is sent back to Deepgram, which converts the text to speech and plays it back to the user in real-time.
• WebSocket Communication: All interactions between the client, Deepgram, and Gemini LLM are handled via WebSocket for low-latency, bi-directional communication.

Tech Stack

• Next.js: React-based framework for building the client-side UI and server-side functionality.
• Deepgram: Speech-to-text API for converting voice to text and text back to voice.
• Gemini LLM: Large language model for generating contextually appropriate responses from user input.
• WebSocket: Enables real-time, full-duplex communication between the client and backend services.

Installation

1. Clone the repository:

   git clone https://github.com/ayushjaiz/interview-platform.git
   
   cd interview-platform

2. Install dependencies:

   npm install

3. Set up environment variables:

   Create a .env file at the root of your project and add your Deepgram and Gemini LLM API keys:

   DEEPGRAM_API_KEY=your-deepgram-api-key
   NEXT_PUBLIC_GEMINI_API_KEY=your-gemini-api-key

4. Start the development server:

   npm run dev

   Your application will run on http://localhost:3000.

Usage

1. Open the application in your browser.
2. Click on the microphone button to start speaking.
3. The application will send your voice to Deepgram for transcription.
4. The transcribed text will be processed by the Gemini LLM.
5. The generated response will be converted back to voice and played for you.

note: Keep your fans closed while running the project