SOP-004: Voice Coding
DOCUMENT CONTROL
| Field | Value |
|---|---|
| SOP ID | SOP-004 |
| Version | 1.0 |
| Status | Active |
| Last Updated | December 2025 |
Purpose
This SOP covers using Happy Coder's voice agent feature for hands-free coding sessions, enabling development while away from your desk.
Voice Agent Architecture
┌─────────────────────────────────────────────────────────────────┐
│ VOICE CODING FLOW │
├─────────────────────────────────────────────────────────────────┤
│ │
│ ┌─────────┐ ┌─────────────┐ ┌─────────────┐ │
│ │ You │───▶│ Voice Agent │───▶│ Claude Code │ │
│ │Speaking │ │ (Translator)│ │ (Coder) │ │
│ └─────────┘ └─────────────┘ └─────────────┘ │
│ │ │ │
│ │ Cleans up your │ Executes │
│ │ rambling into │ structured │
│ │ clear requests │ commands │
│ ▼ ▼ │
│ ┌─────────────────────────────────┐ │
│ │ Your Code Changes │ │
│ └─────────────────────────────────┘ │
│ │
└─────────────────────────────────────────────────────────────────┘Procedure Flowchart
┌────────────────────┐
│ START │
└─────────┬──────────┘
│
▼
┌────────────────────┐
│ Open Happy Coder │
│ mobile app │
└─────────┬──────────┘
│
▼
┌────────────────────┐
│ Enable Voice Agent │
│ in settings │
└─────────┬──────────┘
│
▼
┌────────────────────┐
│ Tap microphone │
│ button │
└─────────┬──────────┘
│
▼
┌────────────────────┐
│ Speak your coding │
│ request naturally │
└─────────┬──────────┘
│
▼
┌────────────────────┐
│ Voice agent │
│ processes & sends │
└─────────┬──────────┘
│
▼
┌────────────────────┐
│ Claude Code │
│ executes task │
└─────────┬──────────┘
│
▼
┌────────────────────┐
│ COMPLETE │
└────────────────────┘Step-by-Step Procedure
Step 1: Enable Voice Agent
- Open Happy Coder app
- Go to Settings
- Toggle Voice Agent ON
- Select preferred language (15+ supported)
Step 2: Configure Voice Settings
| Setting | Recommended Value |
|---|---|
| Speech Recognition | Eleven Labs STT |
| Response Voice | Eleven Labs TTS |
| Auto-Send | After 2s silence |
| Language | Your preference |
Step 3: Using Voice Commands
Tap the microphone button and speak naturally:
Example Voice Commands
- "Add a logout button to the user profile page"
- "Fix the bug in the authentication handler"
- "Create a new API endpoint for user preferences"
- "Refactor this function to use async/await"
Step 4: Voice Agent Processing
The voice agent:
- Transcribes your speech
- Cleans up rambling/backtracking
- Structures the request clearly
- Sends to Claude Code
Best Use Cases
Great for Voice
| Task Type | Why It Works |
|---|---|
| Planning features | Describe ideas naturally |
| Brainstorming | Think out loud |
| Code review feedback | Discuss changes verbally |
| Bug descriptions | Explain issues conversationally |
| Setting up tasks | Queue work while mobile |
Better for Keyboard
| Task Type | Why Keyboard Is Better |
|---|---|
| Detailed code review | Need multiple windows |
| Precise edits | Line-level accuracy |
| Complex git operations | Need to see diff |
| Testing/debugging | Interactive process |
Voice Coding Scenarios
The Commute Pattern
1. Start refactoring task before leaving work
2. Check progress on your commute
3. Give verbal feedback: "Looks good, but move that
function to a separate file"
4. Arrive home to completed workThe Inspiration Pattern
1. Have idea while away from computer
2. Open Happy Coder on phone
3. Voice describe: "I think the race condition is in
the event handler, let's try using a mutex"
4. Claude Code investigates while you continue your dayThe Hammock Pattern
1. Relaxing outside, thinking about code
2. Voice describe approach: "Let's refactor the
Sidekiq job into Temporal activities"
3. Claude Code creates stubs
4. Get excited, go inside to continue on desktop
5. Same session continues seamlesslyVoice Agent Customization
Edit prompts in the mobile app settings:
markdown
You are a friendly, efficient voice translator.
Your job is to take natural speech and convert
it into clear, structured requests for Claude Code.
When I ramble, extract the core request.
When I backtrack, use the final version.
When I'm unclear, ask for clarification.Known Limitations
Current Limitations
- Voice agent costs ~$8/hour (prototype pricing)
- Conversation context may be lost in long sessions
- Best for short, focused interactions
Verification Checklist
- [ ] Voice agent enabled in settings
- [ ] Microphone permissions granted
- [ ] Test voice command works
- [ ] Response plays back correctly
- [ ] Commands reach Claude Code
