Yaar Ki Boli

Where Translations Feel Like a Friend.

Comment

Inspiration

Every time I used AI translators, one thing always bothered me — they only translate words, not emotions. In Indian languages like Hindi and Telugu, there’s a certain “yaari waali vibe”, that friendly tone, which completely disappears in AI translations. That’s when I thought — what if AI translates the sentence, but my dost (friend) adds the real feel to it? That’s how the idea of “Yaar Ki Boli” was born — where AI does the translation, but your friend gives it life.

What it does

“Yaar Ki Boli” is a translator that takes your message and gives you two versions: AI Response — plain and formal translation. Yaar Response (HI) — where your dost replies with emotion, slang, and the typical desi friendship tone.

You’ll instantly feel the difference between a robotic translation and a friend’s reply.

How we built it

This project was built by me and my teammate as a two-person team. I handled the entire backend development, API integrations, emotional tone prompt engineering, and full deployment. My teammate helped in building the frontend interface using React.js, where users can input messages, select languages, and see the AI vs HI responses. I deployed the backend on Render, connected it with the frontend (deployed on Vercel), and made sure the responses feel real, emotional, and friend-like.

Challenges we ran into

The biggest challenge was teaching AI how to “talk like a dost”. Most translation APIs give boring textbook responses. I had to spend a lot of time crafting emotional prompts that bring out that friend-like tone in both Hindi and Telugu. Backend deployment also gave me trouble — CORS issues, environment variables not working — but I figured it out step by step. Another challenge was to keep the responses short, casual, and real, because a real friend always talks straight — no extra formality.

Accomplishments that we're proud of

Built a working demo that clearly shows the difference between AI vs Human-like translations. Designed emotional prompts myself that actually make AI sound like a true dost (friend). Solved all technical hurdles alone — from coding to deployment. Most importantly, created a project that doesn’t just translate words, but also connects hearts.

What we learned

Prompt engineering is everything — a simple API can behave like a human if your prompts are powerful. Language is not just grammar — tone and emotions are equally important. Faced and fixed real deployment problems (CORS, env variables) — it taught me patience. I realized that to build something meaningful, you have to feel the problem yourself first.

What's next for Yaar Ki Boli

Right now, “Yaar Ki Boli” already supports a wide range of Indian languages and dialects — from Khariboli, Braj Bhasha, Bhojpuri, Maithili, Bundeli, Marwari, to Tamil, Telugu, Kannada, Malayalam, and many more. But my journey doesn’t stop here. Next, I want to make sure that every Indian dialect, no matter how small, gets its own “yaar waali” emotional tone. Because in India, language isn’t just about grammar — it’s about how you feel when you say “bhai”, “yaar”, “ra”, “nanna”, “tammudu”. And I want “Yaar Ki Boli” to respect that diversity with full heart.

In the future, I will add Voice Responses, so when you read the translation, it’ll actually feel like your dost is speaking to you, with your own desi vibe and slang.

One day, I want “Yaar Ki Boli” to become more than just a tool — It should be a movement that teaches AI to understand not just words, but real human emotions. Even global platforms like Google Translate should realize that in India, feelings are spoken through dialects, slang, and dosti’s tone.

Built With

Share this project:

Updates