DEV Community: Y H

I Open-Sourced an Android Voice Assistant for OpenClaw

Y H — Fri, 20 Feb 2026 10:55:16 +0000

I’ve been building a practical Android assistant on top of OpenClaw and decided to open-source it.

Repo: https://github.com/yuga-hashimoto/openclaw-assistant

Why I built this

Most assistant demos look good in short clips, but I wanted something usable in daily life:

wake word
voice interaction
reliable connection to agent backends
secure local storage for credentials

What it includes

Offline wake-word detection with Vosk
Android VoiceInteractionService integration (long-press Home)
Real-time OpenClaw chat completions + streaming
Encrypted settings (AES256-GCM) and device identity
Bilingual UI (English/Japanese)

Looking for feedback

If you build assistants, I’d love thoughts on:

setup/onboarding clarity
reliability in long-running usage
what would make this production-ready for teams

Thanks for checking it out 🙌

Introducing OpenClaw Assistant: Open Source Android Voice Assistant with Wake Word Detection

Y H — Fri, 06 Feb 2026 12:25:55 +0000

I'm excited to share OpenClaw Assistant, an open source Android voice assistant that works completely offline!

What is OpenClaw Assistant?

OpenClaw Assistant is a privacy-first voice assistant for Android that provides:

Wake word detection using Vosk (completely offline)
System integration - control your device with voice commands
Text-to-Speech with embedded voices (no internet required)
API integration - connect to your favorite LLM providers
100% open source - fully customizable

Key Features

🎤 Offline Wake Word Detection

Say "Hey Assistant" and it wakes up - no cloud processing, no data leaving your device.

🔒 Privacy First

All speech recognition happens on-device. Your voice data stays on your phone.

🔧 Highly Customizable

Custom wake words
Multiple TTS engines
Plugin system for extending functionality

🤖 AI Integration

Connect to OpenAI, Anthropic Claude, Google Gemini, or any OpenAI-compatible API.

Tech Stack

Kotlin + Jetpack Compose
Vosk for wake word detection
Sherpa-ONNX for embedded TTS
Material 3 Design

Get Started

🔗 GitHub: https://github.com/yuga-hashimoto/OpenClawAssistant

Contributions welcome! Feel free to open issues, submit PRs, or just give it a ⭐

Would love to hear your feedback!

Building an Open Source Android Voice Assistant with Kotlin

Y H — Wed, 04 Feb 2026 03:53:03 +0000

Replace Google Assistant with Your Own AI

What if you could long-press your Home button and talk to YOUR AI instead of Google's?

I built OpenClaw Assistant - an open-source Android app that does exactly that.

📹 Demo: https://x.com/i/status/2017914589938438532

🔗 GitHub: https://github.com/yuga-hashimoto/OpenClawAssistant

Features

🏠 System Assistant Integration - Long-press Home to activate
🎤 Custom Wake Words - "Jarvis", "Computer", or your own
📴 Offline Wake Word Detection - Using Vosk, no cloud needed
🔊 Voice I/O - Speech recognition + TTS
🔗 Any Backend - Connect to Ollama, OpenAI, Claude, or custom APIs

Tech Stack

Component	Technology
UI	Kotlin + Jetpack Compose + Material 3
System Hook	VoiceInteractionService
Wake Word	Vosk (offline)
Speech	Android SpeechRecognizer + TTS
Network	OkHttp + Gson

How It Works

App registers as Android's digital assistant
Vosk listens for wake words locally
On activation, speech is transcribed and sent to your webhook
Response is spoken via TTS

Get Started

git clone https://github.com/yuga-hashimoto/OpenClawAssistant
cd OpenClawAssistant
./gradlew assembleDebug

Or download the APK from Releases.

Backend Setup

Works with OpenClaw or any webhook that accepts:

POST /your-endpoint
{"message": "user's speech", "session_id": "..."}

And returns:

{"response": "AI's reply"}

Contributions welcome! Let me know what you think.

I Built an Experiment Where Two AIs Compete to Build the Best Browser Game

Y H — Fri, 30 Jan 2026 08:15:35 +0000

What happens when you let two AIs build a game from a blank screen?

The Experiment

Self-Evolving Game is an experiment where two AI models (Mimo and Grok) compete to build the most engaging browser game—starting from nothing.

How it works:

Both AIs receive identical instructions and analytics data
GitHub Actions triggers them twice daily (6:00 & 18:00 JST)
Each AI modifies the code, tests pass, auto-deploys
They can add games, fix bugs, improve UX—complete freedom

The result:

In just 2 weeks, they've built:

9+ games (Tetris, Snake, 2048, endless runners...)
Achievement systems
Daily challenges
And more!

Check the changelog to see their "thought process" - each commit includes the AI's intent explaining why they made that change.

🔗 Live site: https://self-evolving.dev/
🔗 Changelog: https://self-evolving.dev/changelogs/compare

Would love to hear your thoughts!