Anthropic has added voice mode to Claude, letting users have spoken back-and-forth conversations through claude.ai and the iOS and Android apps. The beta is live for all plan tiers — Free, Pro, and Team — and is currently English-only.
The feature ships with two modes. Hands-free keeps the microphone open and detects natural speech pauses; push-to-talk hands control back to the user. Either way, you can interrupt Claude mid-response by speaking, and switching between text and voice mid-conversation doesn't break the thread — chat history stays intact throughout.
For those using Claude for research and agentic tasks, the more interesting addition is web search through voice. That combination — hands-free interaction with real-time search — covers situations where typing isn't practical but you still need substantive answers: commutes, cooking, anything that keeps your hands occupied. All voice sessions are transcribed and saved to chat history, which matters for anyone who needs to refer back to what was said.
Anthropichas gone with a curated set of preset voices rather than open-ended voice customization. That's a deliberate call to limit cloning and impersonation risk — an increasingly live concern as synthetic audio becomes more convincing. The English-only scope and beta label both point to further rollout; language expansion seems like the straightforward next move.