VoiceDev — Voice-Native Development

Voice-native development for VS Code — speak to navigate, commit, and control your workflow.

⚠️ v0.1.0-preview — Early development. Expect rapid improvements.

🎥 See It In Action

Watch VoiceDev in action — see voice-driven git workflows, navigation, and AI-assisted development.

Why VoiceDev?

Development is full of small, repetitive actions — saving files, committing code, navigating to a line, running a command. These tiny interruptions pull you out of flow.

VoiceDev makes voice a first-class way to drive your development workflows. Not just dictation — real workflows, entirely by voice:

🗣️ "git status"  →  "git commit message fixed the auth bug"  →  "git push"

A full commit cycle without touching a menu, a palette, or a terminal prompt.

🗣️ "open file server.ts"  →  "go to line 42"  →  "format document"  →  "save all"

Navigate, edit, and save — all spoken.

🗣️ "ask copilot explain this file"  →  "copilot commit"

AI-assisted development, triggered by voice.

Core Beliefs

• Workflows over keystrokes — voice shines at chaining intent, not replacing a keyboard
• Privacy as a feature — cloud speed or local privacy, always your choice
• Accessible by nature — every speakable workflow is one less barrier
• Progressive disclosure — simple on day one, powerful by day thirty

📖 Read the full philosophy in VISION.md

Features

🗣️ 30+ Voice Commands

Speak naturally — VoiceDev understands you even if your wording isn't exact, thanks to fuzzy matching with confidence scoring.

Editor

• "save all" · "format document" · "new terminal" · "close editor"

Git — full workflow by voice

• "git status" · "git add all" · "git diff" · "git log"
• "git commit message fixed the login bug" — wildcard captures your message
• "git pull" · "git push" (with confirmation) · force push blocked for safety

Navigation

• "open file server.ts" — fuzzy file search
• "go to line 42" · "go to top" · "go to bottom" · "go to symbol"

Copilot CLI

• "ask copilot explain this error" · "copilot commit" · "copilot suggest how to list docker containers"

Copilot Chat

• "copilot chat explain this file" · "ask copilot in chat how to fix this" · "open copilot chat"

System

• "help" · "open command center" · "show shortcuts"

🎯 Smart Matching

• Wildcard patterns — say "git commit message fixed the auth bug" and VoiceDev extracts "fixed the auth bug" as the commit message. 9 commands support dynamic argument capture.
• Fuzzy matching — "format the document" still triggers format-document. No need to memorize exact phrases.
• Confidence scoring — if VoiceDev isn't confident enough in a command match, it falls back to dictation instead of guessing wrong.

📝 Voice Dictation

When your speech doesn't match a command, VoiceDev inserts it as text — into the active editor at your cursor, or into the terminal if one is focused.

🔒 Privacy-First Provider Choice

Your voice, your rules. Choose between fast cloud transcription or fully offline local mode:

Provider	Speed	Cost	Privacy	Setup
Groq (default)	⚡ Very Fast	🆓 Free tier	☁️ Cloud	API key only
Voxtral (Mistral)	⚡ Very Fast	🆓 Free tier	☁️ Cloud	API key only
Local	🐌 Slower	🆓 Zero cost	🔒 Fully offline	One-time setup (~2 min)
OpenAI	🐢 Medium	💰 Paid	☁️ Cloud	API key only

Local mode runs via faster-whisper — your voice never leaves your machine. No API keys, no cloud, no limits.

🔊 Audio & Visual Feedback

• Start/stop chimes so you know when VoiceDev is listening
• Status bar shows recording timer, transcription spinner, and provider name
• Toast notifications for command execution, errors, and results
• Command Center webview for browsing all available commands

Roadmap

VoiceDev is in active development. Here's where we are and where we're heading.

🟢 Now (v0.1.0-preview)

• ✅ 30+ voice commands — editor, git, navigation, Copilot CLI, and Copilot Chat
• ✅ Multi-provider speech-to-text (Groq, Mistral Voxtral, OpenAI, Local)
• ✅ Privacy-first offline mode via faster-whisper
• ✅ Fuzzy matching and wildcard pattern extraction
• ✅ Audio feedback and status bar integration

🔵 Next

• 🔗 Action Chaining — compose multi-step voice sequences into reusable workflows. Say "git workflow" to run diff → stage → commit → push in one voice command. Chain any combination of built-in commands with abort-on-failure safety.
• 🎙️ Custom Macros — record a voice phrase and map it to one or more commands. Say "deploy staging" to trigger your own sequence of git pull → build → deploy. Define macros in settings or through the Command Center.
• AI-powered developer workflows (inline completion, coding assists)
• Real-time translation from multiple languages into English
• More VS Code-native actions by voice (search extensions, update settings, open projects)

🟣 Exploring

• Voice-based web search with responses inside the IDE
• Multi-IDE expansion (Open VSX, Zed, and beyond)
• Conversational mode — multi-turn voice interactions
• Developer dictionary — learn your codebase vocabulary
• Full cross-platform production release

📖 See VISION.md for the full philosophy and detailed roadmap

Requirements

• VS Code 1.85.0 or higher
• Microphone for voice input

Cloud Providers (Default)

• Groq API key (free at console.groq.com) - for fast cloud speech-to-text
• OpenRouter API key (optional) - for AI-powered commit messages

Local Provider (Privacy-First Option)

• Python 3.9-3.12 - must be in PATH
• ffmpeg - must be in PATH (download here)
• Disk space: ~150MB for model download (one-time)
• RAM: 4GB minimum, 8GB+ recommended

Why choose local? Your voice never leaves your machine. Zero API costs, unlimited usage, works offline.

Installation

1. Install from VS Code Marketplace (coming soon)
2. Or install the .vsix file manually

Quick Start

Option A: Groq (Fast & Free - Recommended for Getting Started)

1. Get a free API key from console.groq.com
2. Open Command Palette (Cmd/Ctrl + Shift + P)
3. Run VoiceDev: Set API Key, select "Groq", and paste your key
4. Press Ctrl+Shift+V to start recording
5. Speak a command like "save all" or dictate text

Option B: Mistral (Fast & Free - Voxtral powered)

1. Get a free API key from console.mistral.ai
2. Open VS Code Settings (Cmd/Ctrl + ,)
3. Search for "voicedev.stt.provider" and select "mistral"
4. Open Command Palette and run VoiceDev: Set API Key, select "Mistral"
5. Press Ctrl+Shift+V to start recording
6. Speak a command or dictate text

Option C: Local (Private & Offline - Zero Cost)

1. Install Python 3.9-3.12 and ffmpeg (if not already installed)
2. Open VS Code Settings (Cmd/Ctrl + ,)
3. Search for "voicedev.stt.provider"
4. Select "local" from the dropdown
5. Press Ctrl+Shift+V to start recording (first run will setup environment ~2 min)
6. Your voice stays on your machine - no API keys, no cloud, no limits

Configuration

Customize VoiceDev through your VS Code settings. Add these configurations to your settings.json:

Basic Configuration

{
	"voicedev.stt.provider": "groq",
	"voicedev.llm.provider": "openrouter",
	"voicedev.llm.model": "anthropic/claude-3-haiku-20240307",
	"voicedev.audio.feedbackSounds": true
}

Configuration Options

Setting	Description	Values	Default
voicedev.stt.provider	Speech-to-text provider	"groq", "mistral", "openai", "local"	"groq"
voicedev.llm.provider	AI model provider	"openrouter", "openai"	"openrouter"
voicedev.llm.model	AI model for commit messages	Model identifier string	"anthropic/claude-3-haiku-20240307"
voicedev.audio.feedbackSounds	Enable/disable audio feedback	true, false	true

How to Apply Configuration

1. Open VS Code Settings:

   • Windows/Linux: Ctrl + ,
   • macOS: Cmd + ,

2. Click the "Open Settings (JSON)" icon in the top-right corner

3. Add or modify the VoiceDev configuration section

4. Save the file - changes take effect immediately

Provider-Specific Settings

Groq Provider

{
	"voicedev.stt.provider": "groq"
}

Note: Use VoiceDev: Set API Key command to securely store your Groq API key.

Mistral Provider

{
	"voicedev.stt.provider": "mistral"
}

Note: Use VoiceDev: Set API Key command to securely store your Mistral API key. Powered by Mistral's Voxtral-mini model for fast, accurate transcription.

Local Provider

{
	"voicedev.stt.provider": "local",
	"voicedev.stt.local.pythonPath": "path-to-python-executable"
}

OpenAI Provider

{
	"voicedev.stt.provider": "openai"
}

Note: Use VoiceDev: Set API Key command to securely store your OpenAI API key. The OpenAI provider integration is planned for a future release.

For Developers

Interested in contributing to VoiceDev? Check out our QUICKSTART.md guide for detailed setup instructions, development workflow, and information about working with the local/offline speech-to-text provider.

Quick links:

• 📖 Contribution Guidelines
• 🐞 Report Issues
• 💬 Discussions

Reporting Issues

We welcome bug reports and feature requests from both technical contributors and non-technical users! When creating a new issue, you'll be guided through our structured templates:

• 🐛 Bug Report: For reporting bugs and unexpected behavior
• 🚀 Feature Request: For suggesting new features and improvements
• 💬 General Issue: For questions, discussions, or other topics

Tips for effective issue reporting:

• Search existing issues before creating a new one
• Provide clear, step-by-step reproduction instructions for bugs
• Include version information (VoiceDev, VS Code, OS)
• Add screenshots or recordings when helpful
• Be specific about expected vs actual behavior

Our templates include sections for both non-technical users (basic information) and technical contributors (detailed logs, technical insights).

Troubleshooting

Local Provider Issues

"Python 3.9+ is required to use Local STT"

• Ensure Python 3.9-3.12 is installed and in your system PATH
• Run python --version in terminal to verify

"ffmpeg not found in PATH"

• Install ffmpeg from ffmpeg.org
• On Windows: choco install ffmpeg (with Chocolatey)
• On macOS: brew install ffmpeg
• Verify with ffmpeg -version

"Setup failed" or timeout during first run

• Check your internet connection (downloads ~150MB model)
• Run VoiceDev: Clear API Key and try again
• Ensure enough disk space (~200MB total)

Cloud Provider Issues

"Invalid API key" or transcription fails

• Verify your API key at console.groq.com
• Run VoiceDev: Set API Key to update
• Check your internet connection

Known Limitations

Local Provider

• Performance: ~3-5x slower than cloud providers (runs on CPU)
• First run: One-time setup takes ~2 minutes + ~150MB download
• Best for: Short commands and dictation (under 30 seconds)
• Accuracy: Slightly lower with CPU int8 quantization vs full GPU models
• Platform: Windows fully supported; macOS/Linux support in progress

All Providers

• Maximum recording length: 30 seconds (auto-stop)
• Requires clear audio input for best accuracy
• Background noise may affect transcription quality

Contributing

Contributions welcome! See CONTRIBUTING.md for guidelines.

License

BSD 3-Clause License - See LICENSE file for details.

---

Voice-native development — read the vision 🎙️