need the voice chat in code

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

Currently, Claude's mobile app only supports speech-to-text input via the phone's system keyboard dictation. Responses remain plain text and require manually tapping to view — there is no automatic voice playback.

Request 1 — Real-time voice conversation mode:
Please add a continuous voice conversation mode similar to what other AI assistant apps offer: press-and-hold or automatic silence-detection to end a turn, followed by an automatic spoken response — without requiring manual taps for each step. This would make Claude far more usable as a true voice assistant on mobile.

Request 2 — Custom voice cloning:
Please allow users to upload a short audio sample and have Claude's spoken responses read back in that custom cloned voice, instead of being limited to a fixed set of system voices. This would be especially valuable for content creators and users who want a more personalized assistant experience.

Combining these two features would significantly improve Claude's usability on mobile and bring it closer to how people expect a modern voice assistant to work. I believe many users would benefit from this.

Proposed Solution

Add a "Voice Mode" toggle to the Claude Code mobile chat interface, similar to what Claude.ai's voice mode already offers:

1. A microphone button in the chat input bar. Tap and hold to speak, or tap once to start continuous listening with automatic silence detection (e.g. ~1.5s of silence ends the turn).

1. Speech is transcribed and sent as a normal user message into the current conversation — so it has full access to session context, memory, and tools, exactly like a typed message would.

1. Claude's text response is automatically converted to speech and played back, without requiring the user to manually tap to hear it.

1. In Settings, allow the user to upload a short audio sample (e.g. 10s-3min) to clone a custom voice, and select that cloned voice as the one used for spoken responses instead of the default system voices.

1. A visible "end conversation" control to stop the voice session and return to normal text chat.

Alternative Solutions

Currently using the phone's system keyboard dictation to speak text into the chat input, but this requires manual taps for each turn and only handles voice input — there is no automatic spoken reply, so it doesn't provide a true hands-free conversation experience.

Priority

Critical - Blocking my work

Feature Category

CLI commands and flags

Use Case Example

Example scenario:

1. I'm walking or driving and want to ask Claude a question hands-free.
2. I open Claude Code on my phone and tap the voice mode button.
3. I speak my question naturally; Claude transcribes it into the conversation and responds.
4. The response is read aloud automatically in my chosen voice, so I never need to look at or touch the screen.
5. This would let me use Claude the same way I'd use a phone call — fully hands-free, while keeping the full context and tool access of my Claude Code session.

Additional Context

For reference, Claude.ai's own app already has a voice mode with this kind of press-to-talk / auto-detect UX — this request is asking for the same experience inside Claude Code's mobile interface, plus the added custom voice cloning option.