ZenStream

ZenStream is a calming, AI-powered podcast web app that helps users relax, focus, and find serenity. Users can sign in, select or search for topics, and generate personalized podcasts.

Comment

Inspiration

We created ZenStream because we saw how stress and information overload can make it hard for people to relax, focus, or find a sense of calm. Podcasts are a popular way to unwind, but finding truly personalized, uplifting content is difficult. We wanted to harness the power of AI and modern web technologies to deliver calming, positive podcasts tailored to each user’s needs, helping people find moments of serenity in their busy lives.

What it does

ZenStream is an AI-powered web app that generates personalized podcasts to help users relax and focus. Users can securely log in, search or select topics, and instantly receive a podcast script generated by combining relevant inspirational quotes and affirmations. The app reads the script aloud using Text-to-Speech, so users can listen right in their browser. ZenStream is fully responsive and accessible on any device.

How we built it

  • Frontend: Built with Next.js and modern CSS for a clean, responsive UI.
  • Authentication: Clerk provides secure sign-in and user management.
  • Search & Suggestions: As users type topics, ZenStream uses MongoDB Vector Search to suggest relevant ideas from a large dataset of inspirational quotes and affirmations.
  • Podcast Generation: When a topic is selected, the app retrieves relevant quotes and sends them, along with the topic, to the Gemini API from Google Cloud. Gemini generates a coherent podcast script using Retrieval-Augmented Generation (RAG).
  • Text-to-Speech: The podcast script is converted to audio using browser-based Text-to-Speech, so users can listen instantly.
  • Deployment: The app is deployed on Google Cloud for speed and reliability.

Challenges we ran into

  • Ensuring the vector search returned truly relevant and uplifting quotes for a wide range of topics.
  • Integrating Clerk authentication smoothly with the Next.js frontend.
  • Handling browser compatibility for Text-to-Speech, since not all browsers support the Web Speech API equally.
  • Optimizing the RAG pipeline to keep podcast scripts coherent and engaging, even for niche topics.
  • Making the UI intuitive and responsive across all devices.

Accomplishments that we're proud of

  • Seamless integration of Retrieval-Augmented Generation with Gemini API and MongoDB Vector Search, resulting in highly personalized podcast scripts.
  • Creating a smooth, user-friendly experience from sign-in to listening.
  • Deploying a robust, scalable app on Google Cloud that works reliably for all users.
  • Building a tool that genuinely helps people relax, focus, and feel inspired.

What we learned

  • The power of combining semantic search (vector search) with generative AI to create meaningful, context-aware content.
  • The importance of user experience in wellness apps—simplicity and speed matter.
  • How to troubleshoot browser-specific issues with client-side Text-to-Speech.
  • Best practices for deploying and scaling Next.js apps in the cloud.

What's next for ZenStream

  • Expanding the dataset to include more diverse sources of inspiration and positivity.
  • Adding user customization for voice, tone, and podcast length.
  • Introducing features like scheduling, reminders, and sharing.
  • Exploring mobile app versions for offline listening.
  • Continuously improving the AI pipeline for even more natural, engaging podcasts.
Share this project:

Updates