Podcast Accessibility Enhancer transcription and insights application using AssemblyAI Challenge

This is a submission for the AssemblyAI Challenge : Really Rad Real-Time.

What I Built

I created a Podcast Accessibility Enhancer that makes podcasts more accessible and insightful for everyone. Think of it as your smart podcast companion that not only transcribes content but also helps you understand and navigate through it effortlessly. Users can either upload podcast files or paste URLs, and the app does the heavy lifting - creating transcripts, pulling out the important bits, and making everything searchable.

The coolest part? It works in real-time, so you can see the transcription happening as the podcast plays. Plus, it automatically generates chapters, like a table of contents for your ears. I also added features that extract memorable quotes and create automatic hashtags, making it super easy to share specific moments on social media.

Demo

chintanonweb / podcast-accessibility-enhancer

Podcast Accessibility Enhancer

This project aims to enhance the accessibility of podcasts by providing real-time transcription, instant chapter summaries, key insights, searchable transcripts, and automatic hashtag generation.

Table of Contents

• Features
• Installation
• Usage
• Folder Structure
• Environment Variables
• Contributing
• License

Features

• Real-time Transcription: Transcribe podcasts in real-time.
• Instant Chapter Summaries: Generate chapter summaries automatically.
• Key Insights and Quotes: Extract key insights and quotes.
• Searchable Transcript: Provide a searchable transcript with timestamps.
• Automatic Hashtags and Topic Tags: Generate hashtags and topic tags automatically.

Installation

Prerequisites

• Node.js (v14 or higher)
• npm (v7 or higher)
• MongoDB
• Redis
• AssemblyAI API Key

Steps

1. Clone the Repository

   git clone https://github.com/yourusername/podcast-accessibility-enhancer.git
   cd podcast-accessibility-enhancer

2. Install Dependencies

   cd backend
   npm install
   cd ../frontend
   npm install

3. Set Up Environment Variables

   Create a .env file in the backend directory with the following content:

   MONGO_URI=your_mongodb_connection_string
   REDIS_HOST=127.0.0.1
   REDIS_PORT=6379
   ASSEMBLYAI_API_KEY=your_assemblyai_api_key
   JWT_SECRET=

   …

View on GitHub

Journey

The implementation journey was both challenging and exciting. I started with AssemblyAI's Streaming API to handle the real-time transcription. What's neat about their API is how seamlessly it handles different accents and multiple speakers - something crucial for podcast content.

But I wanted to go beyond just transcription, so I tapped into additional AssemblyAI tools:

1. Universal-2 Model: This ensures our transcriptions are spot-on accurate, especially with technical terms and proper nouns that often come up in podcasts.
2. LeMUR: This is where the magic happens. I used it to:
   • Break down long episodes into logical chapters
   • Pull out the most impactful quotes
   • Generate relevant hashtags and topics
   • Create quick summaries of each section

The biggest challenge was making everything work smoothly in real-time while keeping the interface clean and intuitive. I ended up using a combination of websockets for live updates and clever caching to prevent any lag in the user experience.

This project qualifies for multiple prompts as it showcases:

• Sophisticated Speech-to-Text: Using Universal-2 for accurate transcription
• Really Rad Real-Time: Implementing the Streaming API for live transcription
• No More Monkey Business: Leveraging LeMUR for intelligent content analysis

What started as a simple transcription tool evolved into a comprehensive podcast enhancement platform that makes content more accessible, searchable, and shareable for everyone.