Voice Coding on Mac: VS Code + Murmur Setup Guide

TL;DR

Enable hands-free coding: 1) Install Murmur (5 min), 2) Install GitHub Copilot in VS Code, 3) Press Cmd+Space to dictate requirements, 4) Use Copilot Chat to ask AI to code. You're done. Reduce typing by 50-60% while coding in VS Code on macOS.

Why Voice Coding in VS Code?

The Problem VS Code Developers Face

• 8+ hours of typing daily — Leading to RSI and fatigue
• Detailed AI prompts are tedious to type — Copilot Chat requires long, specific prompts
• Context switching — Stopping to think and type breaks flow
• Repetitive comments and documentation — Boilerplate text wastes time

The Voice Coding Solution

Voice dictation + Copilot Chat = Hands-free development:

✓ Speak requirements naturally — Copilot understands conversational prompts better ✓ Faster feature development — AI codes while you review ✓ Reduced typing — 50-60% less keyboard usage ✓ Better focus — Describe what you want, let AI handle syntax ✓ Health benefits — Fewer wrist and hand issues

Prerequisites

Before starting, ensure you have:

• macOS 10.14+ (Intel or Apple Silicon)
• VS Code installed — Free from code.visualstudio.com
• GitHub account — Free, used for Copilot authentication
• USB headset (optional but recommended) — Blue Yeti, Audio-Technica, Plantronics

Step 1: Install Murmur (5 minutes)

Download

1. Visit murmur.ai
2. Click "Download for macOS"
3. File saves to Downloads folder

Install

1. Open Downloads folder
2. Double-click Murmur.dmg
3. Drag Murmur to Applications folder
4. Launch Murmur
5. Grant microphone + accessibility permissions

Configure

1. Click Murmur in menu bar
2. Open Preferences (Cmd+,)
3. Go to "Keyboard Shortcuts"
4. Confirm Cmd+Space is set (or customize)
5. Go to "Audio" and test your microphone

Done. Murmur is ready to dictate in any app, including VS Code.

See detailed installation guide for troubleshooting.

Step 2: Install GitHub Copilot in VS Code (5 minutes)

GitHub Copilot is the AI engine that codes for you. Murmur feeds it voice input.

Install Copilot Extension

1. Open VS Code
2. Go to Extensions (Cmd+Shift+X)
3. Search for "GitHub Copilot"
4. Click "Install" (by GitHub)
5. Reload VS Code (Cmd+R)

Authenticate with GitHub

1. VS Code prompts for GitHub login
2. Click "Sign in with GitHub"
3. Browser opens, sign in with your GitHub account
4. Authorize VS Code access
5. Done — You're authenticated

Verify Installation

1. Open any code file (create a test.js if needed)
2. Copilot suggests code as you type
3. Press Tab to accept a suggestion
4. You should see: Copilot inline completions

If Copilot isn't suggesting code, check:

• You're signed in (Account icon bottom-left)
• Copilot is enabled (Command Palette > "Copilot: Toggle")

Step 3: Open Copilot Chat

Copilot Chat is where voice coding magic happens. You'll dictate requirements here and AI generates code.

Enable Copilot Chat

1. Open VS Code
2. Go to Extensions (Cmd+Shift+X)
3. Search for "GitHub Copilot Chat"
4. Click "Install" (by GitHub)
5. Reload VS Code

Open Chat Panel

Option A: Keyboard shortcut

• Press Cmd+Shift+I — Opens Copilot Chat panel

Option B: Click icon

• Look for chat icon in sidebar (left panel)
• Or: View > Chat

You should see:

• Chat input field at bottom
• Conversation area above
• Ready for questions

Step 4: Your First Voice Coding Session

Now you have everything set up. Time to code with your voice.

Simple Example: React Component

The old way (keyboard):

1. Type: "Create a React component..."
2. Wait for autocomplete/copilot
3. Accept suggestions
4. Type manually for complex parts
5. Time: 10 minutes

The voice way (with Murmur + Copilot Chat):

1. Click in Copilot Chat input
2. Press Cmd+Space (Murmur activates)
3. Say: "Create a React component that displays a todo list with add and delete buttons"
4. Release Cmd+Space (text appears in chat)
5. Press Enter (Copilot generates code)
6. Review code
7. Time: 1 minute

Real Workflow: Build a Feature with Voice

Scenario: You want to build a user profile page component.

Step 1: Open Copilot Chat

Press Cmd+Shift+I to open the chat panel.

Step 2: Dictate Requirements

1. Click in the chat input field
2. Press Cmd+Space (Murmur activates)
3. Speak: "Create a React component called UserProfile that takes a user prop and displays their name, email, avatar, and a bio. Style it with Tailwind CSS and add loading and error states."
4. Release Cmd+Space

What you said: 45 seconds of speaking What you typed: 0 keystrokes

Step 3: Copilot Generates

Press Enter in the chat. Copilot generates:

export const UserProfile = ({ user, loading, error }) => {
  if (loading) return <div className="animate-pulse">Loading...</div>;
  if (error) return <div className="text-red-500">Error: {error}</div>;
 
  return (
    <div className="max-w-md mx-auto p-6 bg-white rounded-lg shadow">
      <img src={user.avatar} alt={user.name} className="w-16 h-16 rounded-full" />
      <h2 className="text-2xl font-bold mt-4">{user.name}</h2>
      <p className="text-gray-600">{user.email}</p>
      <p className="mt-4 text-gray-700">{user.bio}</p>
    </div>
  );
};

Step 4: Iterate by Voice

Want to improve the component? Dictate more refinements:

1. Press Cmd+Space again
2. Say: "Add a button to edit the user profile and one to follow the user"
3. Release Cmd+Space, press Enter
4. Copilot updates the component

No typing required.

Step 5: Accept and Use

1. Copilot shows the updated code
2. Copy the code (Cmd+C)
3. Paste into your component file
4. Done

Total time: 3 minutes vs. 20 minutes of manual coding

Advanced Workflow: Full Feature with Voice

Example: Building a Todo App

What you want: Full todo app with add, delete, complete functionality.

The voice approach:

1. [Cmd+Space] "Create a React todo app component. It should display a list of todos, each with name and completion status. Add inputs for creating new todos and buttons for delete and toggle complete."

[Copilot generates complete Todo component]

2. [Cmd+Space] "The todos should be stored in state using useState. Add a useEffect that loads todos from localStorage on mount."

[Copilot adds state management and localStorage]

3. [Cmd+Space] "Add a button to clear all completed todos. Style it with Tailwind and add a count of remaining todos."

[Copilot refines the UI]

4. Done. Full functioning app built entirely by voice.

Comparison:

Pro Tips for Voice Coding in VS Code

Tip 1: Be Descriptive, Not Prescriptive

Bad (prescriptive): "Create a function that iterates through an array using forEach..."

Good (descriptive): "Create a function that processes an array of user objects and returns their names sorted alphabetically"

AI is better at understanding what you want, not how to code it.

Tip 2: Include Context in Your Prompt

Add framework, language, and pattern context:

Good: "Create a React component using hooks that fetches data from an API endpoint"

Better: "Create a React component using useState and useEffect that fetches user data from /api/users and displays it in a table with loading and error states"

More context = better code generation.

Tip 3: Use Separate Prompts for Complex Features

Instead of one giant prompt:

❌ Don't: "Create a complete authentication system with login, signup, password reset, email verification, and two-factor..."

✓ Do:

1. "Create a login form component"
2. "Add password validation logic"
3. "Create a signup form"
4. "Add password reset email flow"

Multiple prompts = better results.

Tip 4: Spell Out Technical Terms

When dictating unusual variable names or domain-specific terms:

"Create a function called L-O-A-D underscore state underscore from underscore cache"

This prevents transcription errors on uncommon words.

Tip 5: Review AI-Generated Code

Always review code before accepting it:

• Is it correct? (Test it)
• Does it match your intent?
• Are there security issues?
• Is it efficient?

AI-generated code is usually excellent, but spot-checking is good practice.

Tip 6: Use Voice for Comments and Documentation

// [Cmd+Space]
// Say: "Function that validates email addresses using regex and returns true if valid"
// [Release]
 
function validateEmail(email) {
  const regex = /^[^\s@]+@[^\s@]+\.[^\s@]+$/;
  return regex.test(email);
}

Voice is perfect for:

• Function documentation
• Inline comments
• Docstrings
• README updates

Tip 7: Combine Voice and Keyboard Strategically

Use voice for:

• Long prompts to Copilot Chat
• Detailed comments and documentation
• Feature descriptions
• Requirements and specifications

Use keyboard for:

• Quick edits (remove extra line, fix typo)
• Debugging and testing
• Precise modifications
• Quick variable name changes

Example workflow:

1. [Voice] Dictate feature to Copilot → AI codes
2. [Keyboard] Review code, make small edits
3. [Voice] Dictate refinement → AI improves
4. [Keyboard] Final tweaks and tests

Keyboard Shortcuts for Voice Coding in VS Code

Essential Shortcuts

Voice Coding Workflow Shortcuts

WORKFLOW:

1. [Cmd+Shift+I]     Open Copilot Chat
2. [Cmd+Space]       Start Murmur dictation
   [Speak requirement]
   [Release Cmd+Space]
3. [Enter]           Submit to Copilot
4. [Review code]     Read the generated code
5. [Cmd+C]           Copy code to clipboard
6. [Cmd+P]           Quick open file
7. [Cmd+V]           Paste code into file
8. [Repeat]          More features, more voice

Troubleshooting Common Issues

Issue: Murmur Cmd+Space Conflicts with Spotlight

Solution:

1. System Preferences > Keyboard > Shortcuts
2. Spotlight > Uncheck "Show Spotlight search"
3. Or change Murmur shortcut to Cmd+Shift+Space

Issue: Copilot Chat Not Appearing

Solution:

1. Command Palette (Cmd+Shift+P)
2. Type "Copilot Chat: Focus Chat"
3. Or install "GitHub Copilot Chat" extension

Issue: Copilot Not Suggesting Code

Solution:

1. Check sign-in: Account icon (bottom-left) should show you're authenticated
2. Enable Copilot: Command Palette > "Copilot: Toggle"
3. Reload VS Code (Cmd+R)

Issue: Poor Microphone Quality

Solution:

1. Test in Murmur settings (Murmur > Preferences > Audio)
2. Move microphone 6-12 inches from mouth
3. Use a better headset (USB headset recommended)
4. Minimize background noise

Daily Voice Coding Schedule

Morning (30 minutes)

1. Open VS Code
2. Open Copilot Chat (Cmd+Shift+I)
3. [Cmd+Space] Describe first feature
4. [Enter] Copilot codes
5. Review and integrate code

Midday (4 hours)

• Use voice for complex features, documentation
• Use keyboard for testing, debugging, small edits
• Alternate between voice prompts and code review

Afternoon (2 hours)

• Document code by voice
• Create comments explaining complex logic
• Write README updates using voice

Evening (30 minutes)

• Write commit messages by voice
• Document tomorrow's tasks by voice
• Stop coding and rest your hands

Result: 50-60% less typing, same or better productivity

Integration with GitHub

Commit Messages by Voice

git add .
 
# [Cmd+Space]
# Say: "Add user profile component with loading and error states"
# [Release]
 
git commit -m "Add user profile component with loading and error states"

Pull Request Descriptions by Voice

1. Create PR on GitHub
2. Click "Write description"
3. [Cmd+Space] Dictate description
4. [Release] Text appears
5. Submit PR

No typing required for documentation.

Extending Voice Coding

Copilot in Terminal

You can even use voice in VS Code's integrated terminal:

# [Cmd+Space]
# Say: "List all JavaScript files in the src directory"
# [Release]
 
ls src/*.js
 
# Run whatever command you dictated

Voice Code Review

Review PRs by voice:

1. Open PR in VS Code
2. Click "Comment" button
3. [Cmd+Space] Dictate feedback
4. [Release] Comment appears
5. Submit review

Best Practices for Voice Coding on Mac

Do

✓ Use descriptive, natural language prompts ✓ Break complex features into multiple voice prompts ✓ Review AI-generated code before accepting ✓ Document your voice prompts (they become git history) ✓ Take breaks (voice fatigue is real) ✓ Use a good quality USB headset ✓ Combine voice and keyboard strategically

Don't

✗ Try to dictate exact syntax (let AI handle it) ✗ Use one giant prompt for complex features ✗ Blindly accept all AI suggestions (always review) ✗ Dictate continuously (take 5-minute breaks) ✗ Dictate while tired (accuracy suffers) ✗ Use poor quality built-in mics (USB headsets are worth it)

Conclusion: Voice-First Development in VS Code

With Murmur + Copilot Chat in VS Code on macOS, you can:

✓ Code entirely hands-free ✓ Reduce typing by 50-60% ✓ Let AI handle syntax while you focus on logic ✓ Protect your hands from RSI ✓ Improve code quality through natural language prompts ✓ Develop faster with less fatigue

Getting started is simple:

1. Download Murmur (5 min)
2. Install Copilot + Copilot Chat (5 min)
3. Press Cmd+Space and start dictating (immediate)

You're now ready for voice-first development in VS Code.

Ready to code with your voice? Download Murmur today and set up voice coding in VS Code. Your hands will thank you.

Related articles:

• How to Dictate Code with Murmur
• Voice Typing for Mac Developers
• Murmur: Voice Dictation for Developers