Voice-to-Action App

Author: Yasharth Bajpai

Date: February 21, 2025

---

Project Overview

The Voice-to-Action App is a React Native application that allows users to record audio, transcribe it into text, and extract actionable insights such as meeting details, key points, decisions, and follow-up tasks. The app integrates with Google Speech-to-Text API for transcription and Perplexity AI for advanced analysis of the transcript.

This project is designed to streamline meeting management by providing users with:

• A transcript of the recording.
• Structured meeting summaries.
• Actionable to-do lists.
• Email integration for sharing meeting details.

---

Features

Audio Recording

• Record audio directly within the app using the device's microphone.
• Supports both iOS and Android platforms.

Speech-to-Text Transcription

• Converts recorded audio into text using the Google Speech-to-Text API.

AI-Powered Insights

• Extracts structured data from transcripts using Perplexity AI, including:
  • Meeting details (date, time, participants).
  • Key discussion points.
  • Decisions made.
  • Action items with assignees and deadlines.

Email Sharing

• Compose and send meeting summaries via email directly from the app.

Editable Meeting Details

• Modify meeting details and tasks through an intuitive edit modal.

Dark-Themed UI

• A visually appealing dark mode design for better user experience.

---

Demo Video

Here is my main link where I have helped people understand how my project works by giving a hands-on demo:
Watch Demo

---

Code Walkthrough

Here is my second link where I have explained how the code works in the backend and frontend, line by line:
Watch Code Walkthrough

---

Technologies Used

Frontend

• React Native
• Expo AV (for audio recording)
• Axios (for backend communication)
• React Native Modal (for modals)
• React Native Linking (for email integration)

Backend

• Node.js with Express.js
• Google Speech-to-Text API
• Perplexity AI API

---

Installation Guide

Prerequisites

1. Install Node.js and npm.
2. Install Expo CLI for running the React Native app.
3. Set up a Google Cloud account and enable the Speech-to-Text API.
4. Obtain an API key from Perplexity AI.

---

Steps

Backend Setup

1. Clone the repository and navigate to the backend folder:

   git clone 
   cd backend
   

2. Install dependencies:

   npm install
   

3. Add your Google Cloud credentials JSON file (myjson.json) to the backend folder.
4. Replace perplexityApikey in transcriptionService.js with your Perplexity AI API key.
5. Start the server:

   node index.js
   

Frontend Setup

1. Navigate to the frontend folder:

   cd frontend
   

2. Install dependencies:

   npm install
   

3. Update the myip variable in App.js with your local machine's IP address (e.g., 192.168.x.x).
4. Start the Expo development server:

   expo start
   

5. Scan the QR code from Expo on your mobile device or run it on an emulator.

---

Usage Instructions

1. Launch the app on your mobile device or emulator.
2. Tap "🎤 Start Recording" to begin recording audio.
3. Tap "⏹️ Stop Recording" to stop recording and process the audio.
4. View the generated transcript and structured insights (meeting details, tasks, etc.).
5. Edit any details by tapping the edit icon ✏️.
6. Share meeting summaries via email by tapping the email icon 📧.

---

Folder Structure

project/
├── backend/
│   ├── index.js                # Main server file
│   ├── transcriptionService.js # Handles transcription and AI analysis
│   └── myjson.json             # Google Cloud credentials file (not included in repo)
├── frontend/
│   ├── App.js                  # Main React Native app file
│   ├── styles.js               # Styling for UI components
│   └── assets/                 # App assets (if any)
└── README.md                   # Documentation file

Screenshots

Home Screen	Transcript & Meeting Details

---

Future Enhancements

1. Add support for multiple languages in transcription.
2. Integrate calendar APIs (e.g., Google Calendar) for automatic event creation.
3. Enable offline transcription using local models.

---

Contact

For any questions or feedback, feel free to reach out:
Yasharth Bajpai
Email: yasharthbajpai0103@gmail.com

---