VibeTalk – Speak English, Match by Mindset

🚀 Inspiration

In India, millions are trying to speak English fluently — not just to pass exams, but to access opportunities, express their ideas, and connect globally. While apps like Duolingo or YouTube tutorials offer grammar lessons, most people drop out because of a deeper issue: they don’t have anyone to speak with comfortably.

The root cause isn’t lack of vocabulary — it’s lack of confidence, motivation, and relatable conversation partners. That’s when the idea struck:

“What if English practice felt like a natural conversation between two like-minded people?”

That’s how VibeTalk was born — an app that matches users based on interests and mindset, and gets them talking in English through fun, judgment-free conversations.


💡 What it does

VibeTalk is a speaking-first English fluency app that:

  • 🎯 Lets users select interests (tech, music, memes, startups, etc.)
  • 🤖 Uses AI to match them with like-minded partners
  • 🎙️ Connects them in 1:1 voice rooms for 5–15 mins
  • 💬 Gives pop-up conversation prompts to spark discussion
  • 📊 Tracks user fluency over time based on speech patterns
  • 🏅 Adds gamified features like coins, streaks, and fluency ranks
  • 🛡️ Offers anonymous voice mode for shy speakers

Rather than just “learning English,” users are vibing in English — and that makes all the difference.


🛠️ How we built it

We designed the MVP with simplicity and scalability in mind:

  • Frontend: Flutter (for Android/iOS compatibility)
  • Backend: Firebase (for auth, Firestore, Realtime Database)
  • Voice Rooms: Agora SDK (for low-latency audio)
  • Matching Engine: Tag-based interest matching (Phase 1), planned upgrade to sentence embedding matching using BERT or SentenceTransformers
  • Fluency Scoring: Speech-to-text with analysis on speed, clarity, and tone confidence
  • Gamification: Firebase triggers to manage coins, streaks, and leaderboards

⚔️ Challenges we ran into

  1. Real-time matching: Ensuring both users are online and ready to talk at the same time without delays.
  2. User anxiety: Many users hesitate to speak. We built features like anonymous voice-only chats and blur profiles to reduce fear.
  3. Interest matching accuracy: Tag-based matching felt shallow at times. We're exploring more nuanced matching using AI-powered vibe graphs.
  4. Retention: Keeping users coming back daily was tough — gamification and fluency graphs helped improve this.
  5. Voice room stability: Managing smooth voice calls across variable networks required optimizing audio SDK settings.

🏆 Accomplishments that we're proud of

  • 🔥 Created a working MVP with real-time voice rooms and interest-based matching
  • 🎉 Helped shy users speak English more confidently for the first time
  • 📊 Built a prototype fluency tracker that scores conversations by speech clarity
  • 🧠 Came up with a concept that blends AI + emotion + language learning — a new twist in the edtech space

📚 What we learned

  • Language learning is emotional. Comfort, safety, and shared vibe matter more than perfect grammar.
  • AI can power empathy. Even a simple interest-based matching system makes users feel more at ease.
  • You don’t need 100 features. A focused MVP that solves the real problem can make users stick.
  • Feedback loops matter. Fluency tracking and visual progress graphs give users a reason to return.

🔮 What's next for VibeTalk

  • 💡 Upgrade AI matching with vibe embeddings + user feedback loops
  • 👥 Launch interest-based clubs for group conversations
  • 📱 Build a clean Figma UI and launch the Android beta
  • 🧑‍🎓 Partner with colleges to launch “English Speaking Circles”
  • 🎓 Offer fluency certificates and partner with job platforms
  • 💰 Add monetization with premium plans and daily challenge boosters

VibeTalk isn’t just an app — it’s a comfort zone where fluency flows naturally.

Let’s help Gen Z find their English voice — one real conversation at a time.

Share this project:

Updates

Submission history