Skip to content

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

Vaaya Flow 🎤

Dear Wispr Flow Team: I'm sorry. I loved your keyboard so much I basically made a knockoff. But hey, imitation is the sincerest form of flattery, right? Please don't sue me. I'm just a humble developer who wanted voice typing that doesn't make us sound like robots. ❤️


Preview

Vaaya Flow Demo


What is this?

Vaaya Flow is an Android voice-to-text keyboard that actually understands you're a human being with things to say.

You know that feeling when you use voice typing and it gives you a wall of text with no punctuation like some kind of stream of consciousness poetry from a caffeinated octopus? Yeah, I fixed that.

The Problem

Standard voice keyboards:

hey can you send me the report by friday also dont forget to include the budget numbers thanks

The Vaaya Flow Solution

Hey, can you send me the report by Friday? Also, don't forget to include the budget numbers. Thanks!

Magic? No. AI? Yes. Hotel? Trivago.


Features That'll Make You Say "Finally"

  • 🎯 Real-time transcription using Deepgram (because waiting is for dialup modems)
  • 🤖 Smart formatting with OpenAI GPT-4o-mini (punctuation, capitalization, the works)
  • 🔒 BYOK (Bring Your Own Keys) - Your API keys, your data, your business
  • ⚡ Streaming responses - See your words appear like you're in a sci-fi movie
  • 🎨 Command mode - Tell it to "formalize this" or "make it funny" and watch the magic happen
  • 🎙️ Push-to-talk with Volume Down - Hold the volume down button for the best experience!

How It Works

💡 Pro Tip: Hold the volume down button while speaking for the best experience!

  1. You talk into your phone like a normal human (hold volume down button for best results!)
  2. Deepgram turns your speech into text in real-time (via WebSocket because I'm fancy)
  3. OpenAI cleans it up and makes you sound professional (or funny, or whatever you want)
  4. Text appears in whatever app you're using
  5. You feel like you're living in the future

Architecture (For the Nerds)

I went full Clean Architecture on this because I have standards:

Domain Layer (Pure Kotlin, no Android)
    ↓
Data Layer (API clients, repositories)
    ↓
Presentation Layer (UI, ViewModels)
    ↓
IME Layer (The keyboard itself)

Tech Stack:

  • Kotlin (obviously)
  • Hilt for DI (because manually managing dependencies is for masochists)
  • Coroutines + Flow (async is the future)
  • MVVM (I'm not an animal)
  • WebSockets for that real-time goodness

Getting Started

Prerequisites

You'll need:

  • Android Studio (the latest one, don't be that person running an ancient version)
  • A Deepgram API key (get one here)
  • An OpenAI API key (get one here)
  • Basic knowledge of Kotlin (or the willingness to learn)
  • A device running Android 8.0+ (Min SDK 26)

Building

# Clone this bad boy
git clone https://github.com/MaruPelkar/vaaya.git
cd vaaya

# Build debug APK
./gradlew assembleDebug

# Install on your device
./gradlew installDebug

# Or build release (for the brave)
./gradlew assembleRelease

Setup

  1. Install the app
  2. Go to Settings → System → Languages & Input → On-screen keyboard
  3. Enable Vaaya Flow
  4. Open the app and add your API keys (Settings icon in the keyboard)
  5. Start typing... with your voice!
  6. 🔊 Best Experience: Hold the volume down button while speaking!

Why BYOK?

I'm not running a backend (yet). You bring your own API keys because:

  1. Privacy: Your keys, your data. I never see it.
  2. Cost: Pay for what you use. No subscriptions, no sneaky charges.
  3. Flexibility: Swap providers whenever you want.
  4. Transparency: You know exactly where your data goes.

Think of it like BYOB, but for APIs.


Configuration

The keyboard is smart, but not telepathic (yet). You can configure:

  • Deepgram API Key: Powers the speech recognition
  • OpenAI API Key: Handles the formatting magic
  • Formatting Preferences: (Coming soon™)

Project Structure

app/src/main/java/com/vaaya/flow/
├── ime/                          # The keyboard itself
│   └── VoiceInputMethodService.kt
├── domain/                       # Business logic (pure Kotlin)
│   ├── model/
│   ├── repository/
│   └── usecase/
├── data/                         # Data layer
│   ├── remote/
│   │   ├── deepgram/            # WebSocket streaming
│   │   └── openai/              # GPT formatting
│   └── repository/
├── presentation/                 # UI layer
│   └── settings/
└── di/                          # Dependency Injection

Want more details? Check out CLAUDE.md (yes, I have a file specifically for AI pair programming).


Contributing

Found a bug? Want to add a feature? PRs are welcome!

Please:

  • Write tests (I'm trying to be professional here)
  • Follow the existing code style (Kotlin conventions)
  • Update documentation if needed
  • Be nice in code reviews

Development Workflow

  1. Fork the repo
  2. Create a feature branch (git checkout -b feature/amazing-thing)
  3. Make your changes
  4. Test everything (seriously, test it)
  5. Commit with meaningful messages (no "fix stuff" commits please)
  6. Push and open a PR

Roadmap

Things I'm working on (in no particular order):

  • Offline mode (because internet isn't everywhere)
  • Multi-language support (English is not the only language)
  • Custom commands (make your own voice shortcuts)
  • Voice editing (say "delete that" and watch it happen)
  • Haptic feedback (because pressing buttons is satisfying)
  • Better error handling (currently I just panic)
  • Dark mode UI improvements
  • Widget support
  • Maybe make peace with the Wispr Flow team?

Known Issues

  • Requires internet (I'm working on offline mode)
  • English only for now (multi-language coming soon)
  • Occasionally gets confused by heavy accents (AI is learning)
  • Can't read your mind (yet)

FAQ

Q: Is this free? A: The app is free. API usage costs depend on Deepgram and OpenAI pricing. Check their websites for details.

Q: Does this work offline? A: Not yet. I'm working on it!

Q: Can I use other AI providers? A: Not currently, but the architecture is designed to make this easy to add. PRs welcome!

Q: Why is it called Vaaya? A: In Hindi, "Vaani" means voice. I made it sound cooler. Marketing 101.

Q: Are you actually sorry about copying Wispr Flow? A: Deeply. But also, competition breeds innovation, so maybe they'll make something even better now?

Q: What's the best way to use the keyboard? A: Hold the volume down button while speaking! This gives you the best push-to-talk experience.


License

MIT License - See LICENSE for details.

Translation: Do whatever you want with this code. Build something cool. Make money. Just don't sue me.


Credits

Made with ☕ and questionable life choices by MaruPelkar - a developer who was tired of punctuating voice messages manually.

Special thanks to:

  • Wispr Flow for the inspiration (and I'm genuinely sorry)
  • Deepgram for incredible real-time transcription
  • OpenAI for making text formatting magical
  • You for reading this far down the README

Stay in Touch


If this saved you from manual punctuation hell, give it a ⭐

Your voice deserves better formatting.

🎙️ Remember: Hold volume down for the best experience!

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages