Claude Code just introduced native voice mode — a built-in feature that lets you speak directly to your terminal instead of typing. No extensions. No workarounds. Just hold spacebar and talk.
This guide covers everything you need to know: what it is, when you're getting it, and how to use it in practice.
What Is Claude Code Voice Mode?
Voice mode is a native feature inside Claude Code that lets you interact with your AI coding assistant using your voice. Instead of typing out long prompts, you hold your spacebar, speak your instruction, and release. Your words stream directly into your prompt — like push-to-talk for your terminal.
It was announced on March 3, 2026 by Anthropic engineer Thariq Shihipar.
When Is It Coming Out?
How to check if you have access: Open Claude Code. If voice mode is available to you, you'll see a note on the welcome screen. If it's there — you're in.
How to Enable Voice Mode
Step 1: Open Claude Code
Launch your terminal and start Claude Code as you normally would.
Step 2: Toggle Voice Mode On
Type the slash command:
/voice
This toggles voice mode on (or off if it's already active).
Step 3: Hold Spacebar and Talk
Once voice mode is active:
- Hold spacebar — this activates push-to-talk
- Speak your command — say what you want Claude to do
- Release spacebar — your words stream into the prompt and Claude executes
That's it. No setup files. No config. No API keys.
What to Expect
Key things to know before you start:
- Voice mode uses push-to-talk (hold spacebar to speak, release to send)
- Your transcript streams in at your cursor position — you can type half a prompt, voice the middle, and keep typing
- It doesn't replace what's already in your prompt — it adds to it
- Voice transcription tokens are free and don't count against rate limits
- Works on Pro, Max, Team, and Enterprise plans
Practical Use Cases
Here's where voice mode actually shines in a real workflow:
1. Explain Your Codebase Hands-Free
Hold spacebar and say:
"What does my project do?"
Claude reads your entire codebase and explains it back to you. Nothing gets modified. Great for onboarding or quickly re-orienting yourself on an old project.
2. Refactor Without Typing
Hold spacebar and say:
"Refactor the authentication middleware to use JWT tokens instead of session cookies."
Claude understands the instruction and executes the refactor. No more typing out long, detailed prompts.
3. Debug Conversationally
Hold spacebar and say:
"Why is the API returning a 403 on the /users endpoint?"
Claude investigates your code, checks the auth logic, and explains the issue — all from a spoken question.
4. Quick File Operations
Hold spacebar and say:
"Create a new utility file called formatDate.ts with a function that converts ISO strings to readable dates."
Faster than typing, especially for repetitive scaffolding tasks.
5. Hybrid Typing + Voice
Start typing a prompt, then hold spacebar to voice the messy middle part, then keep typing. The transcript drops right where your cursor is. Perfect for when you know the structure but want to freestyle the details.
Pro Tips
Start simple. Your first voice command should be read-only — something like "What does this file do?" or "Summarize my project structure." This lets you see how it works without any risk of unwanted changes.
Be specific. Voice mode transcribes exactly what you say. Speak clearly and include file names, function names, or paths when relevant. The more precise you are, the better Claude performs.
Use it for context dumps. If you have a complex idea that's hard to type out — a multi-step refactor, a new feature spec, a debugging theory — just talk it out. Voice is faster than typing for unstructured thoughts.
Pair with slash commands. Use /voice to toggle on, then combine with other slash commands in your workflow. Voice mode doesn't disable any existing Claude Code features.
Quick Reference Cheat Sheet
Want more Claude Code guides? Check out the full library of setup guides, cheat sheets, and skill tutorials in the Lead Magnet Hub.
Disclaimer: This guide is based on publicly available information as of March 2026. Voice mode is currently in a phased rollout. Features and availability may change as Anthropic expands access. Always refer to official Anthropic documentation for the latest updates.