Beginner

Voice & STT/TTS

Claude supports voice input and voice output across its apps. You can speak your prompts and hear Claude's responses read back aloud — useful for hands-free use, accessibility, long responses you want to listen to rather than read, or voice-first workflows on mobile.

Voice Mode — Claude.ai

Voice mode launched on iOS and Android in May 2025 and becamefree for all plan tiers (including Free) on June 3, 2025. It was rolled out to the web and desktop apps throughout the second half of 2025.

How it works

• Tap the microphone icon in Claude.ai to start voice input
• Speak your prompt — Claude transcribes it in real time
• Claude responds in text and reads the response aloud
• Interrupt at any time to change the prompt

Voice selection

Five voices to choose from, each with distinct character:

• Buttery — warm, smooth
• Airy — light, conversational
• Mellow — calm, relaxed
• Glassy — clear, crisp
• Rounded — full, balanced

TTS powered by ElevenLabs technology

Language Support

Voice mode launched with English as the primary language. Broader language support is progressively rolling out through 2026. Claude's underlying text understanding is multilingual, so you can type in many languages — but spoken voice input may have limited accuracy for non-English languages depending on your region and app version.

Voice in Claude Code

Claude Code rolled out a /voice command on March 3, 2026. Toggle it within a Claude Code session to switch to voice mode:

# In a Claude Code session, toggle voice on/off:
/voice

# Now speak your coding prompts — Claude Code listens,
# responds via audio, and continues executing tools as normal

Voice in Claude Code is designed for hands-free coding — speak instructions while keeping your hands on the keyboard for other tasks. It was rolling out gradually, starting with approximately 5% of users in early March 2026.

Upcoming: Offline Voice

Anthropic announced offline voice packs for Q1/Q2 2026. These enable on-device voice processing for prompts up to 30 seconds without an internet connection. Targeted at enterprise customers with strict data residency requirements and educational environments with unreliable connectivity.

Enterprise: Custom Voice Cloning

On the 2026 roadmap for Claude Enterprise: organisations will be able to clone their spokesperson's or internal brand voice (with opt-in privacy controls). This lets companies have Claude respond in a voice that matches their brand identity — useful for customer-facing deployments.

When to Use Voice Mode

Good for voice

• Commuting — ask questions while walking/driving
• Long responses — listen instead of read
• Hands-free coding or note-taking
• Accessibility — faster than typing for many users
• Quick questions on mobile

Stick to text for

• Complex technical prompts with code, URLs, or exact syntax
• Multi-step instructions with specific formatting
• Reviewing output you need to copy-paste
• Non-English languages (until fuller support ships)

Checklist: Do You Understand This?

When did voice mode become free for all Claude.ai users?
What are the five voice options and how do you select them?
How do you enable voice mode in Claude Code?
What is the planned offline voice feature for enterprise?