I Built a Voice AI App and Gave It Away Free — Here's Why

I Paid Rs.3,800 for Something I Could Build Myself

It started with frustration.

I was building AI products — full-stack apps with FastAPI, Next.js, Supabase, LangGraph. The kind of work where you write code for 2 hours and then spend another 3 hours typing Slack messages, commit messages, documentation, PR reviews, and emails.

I think at 150 WPM. I type at 45 WPM. There is a constant bottleneck between what I want to say and what my fingers can produce.

So I bought Wispr Flow. Rs.3,800 per year. A Mac app that lets you speak and it types for you. Everywhere. In VS Code, in Slack, in Chrome. It is genuinely good.

But then I looked at my students.

The Rs.3,800 Problem

I teach AI engineering to 500+ students. Most of them are in India. For an Indian student or early-career developer, Rs.3,800 is not nothing — that is a month of internet, or two weeks of food.

When I told my class about voice typing, the first question was always: "Is there a free version?"

The answer was no. Wispr is $50/month internationally. BridgeVoice — another competitor — also $50/month. Apple's built-in dictation is decent but cannot auto-type into VS Code or Terminal.

There was no good free option. So I built one.

One Night, One App

I am not going to pretend this was a heroic 6-month engineering effort. It was not.

The building blocks already exist:

• WhisperKit by Argmax — runs OpenAI's Whisper model on-device on Apple Silicon
• Swift — Apple's native language with full macOS access
• CGEvent — Apple's API for simulating keyboard events (auto-type)

The hard part was not the technology. It was the polish:

• Making the menu bar app feel native
• Getting accessibility permissions to work reliably
• Adding smart punctuation ("comma" types a comma)
• Voice commands ("scratch that" to undo)
• Supporting 5 different Whisper model sizes
• Building an LLM cleanup layer with 7 providers
• Making it work in EVERY app — VS Code, Terminal, Slack, Chrome, Notes

14 Swift files. 2MB download. One night.

Why Free?

People ask me this a lot. "You could charge Rs.500/month and make good money."

Here is my thinking:

1. Voice input should be infrastructure, not a product.

Typing is free. Your keyboard does not charge you per keystroke. Voice-to-text is just another input method — it should be equally free.

2. Free tools build distribution.

I am not trying to build a voice AI company. I am an AI engineer who builds products for clients. Every student who uses IndianWhisper knows my name. Every LinkedIn post about it reaches 10,000+ people. That is marketing I could not buy.

3. The real money is in what comes next.

IndianWhisper is a tool. The skills I demonstrated building it — shipping production Swift apps, Next.js websites, Chrome extensions, auto-update systems — those skills are worth Rs.3 lakh/month to the right client.

4. Open source compounds.

When you give something away, people contribute. They find bugs. They suggest features. They share it with their network. One free tool creates more value than a hundred paid-but-unused products.

What I Actually Shipped

In one session, here is what got built:

Mac App (Swift):

• 14 source files, WhisperKit 0.9.0+
• 5 Whisper models (75MB to 3GB)
• Auto-type via CGEvent — works in any app
• Smart punctuation and voice commands
• Hindi/Hinglish support
• 7-provider LLM text cleanup
• Auto-update system with version checking
• 2MB DMG, macOS 14+

Website (Next.js 16):

• Premium dark UI with animated light streaks
• Live voice demo (browser, no install)
• ROI calculator showing time and money saved
• SEO-optimized blog posts
• Feedback form with voice input
• Model comparison table
• Download with Gatekeeper bypass instructions

Chrome Extension:

• Floating mic button on every webpage
• Works on Gmail, Slack, Docs, LinkedIn — any text field
• Smart punctuation support
• Keyboard shortcut (Ctrl+Shift+S)
• Under 50KB, zero performance impact
• Submitted to Chrome Web Store

The Response

I posted on LinkedIn: "I paid Rs. 3,800 for a voice typing app. Then I built a better one. For free."

Within 48 hours:

• People started downloading and testing
• Comments asking for Windows and Android versions
• Sales teams started using it
• Students in my bootcamp adopted it

The demand validated what I already knew: people want voice input, they just do not want to pay $600/year for it.

What Is Next

The roadmap is clear:

1. Chrome Extension — submitted, launching this week. Works on every OS.

2. Windows — native app for developers on Windows

3. Android — APK download, no Play Store needed

4. iOS — PWA or App Store when it makes sense

The Mac app plus Chrome extension covers 90% of use cases. If you can open Chrome, you can use IndianWhisper. On any device. For free.

The Lesson

Do not compete on features. Compete on access.

Wispr is a more polished product than IndianWhisper. They have a dedicated team, years of refinement, better error correction. I am not going to pretend otherwise.

But Wispr costs $600/year. IndianWhisper costs $0. For a student in Bangalore, for a freelancer in Lagos, for a developer in Jakarta — that difference is everything.

The best product is not always the most polished. Sometimes it is the one that is available.

Try It

• Website: indianwhisper.com
• Mac Download: Direct DMG
• Chrome Extension: Coming this week
• Live Demo: Try in browser — no install needed
• GitHub: Open source

Your voice is faster than your fingers. Stop paying for that privilege.