[BUG] voiceEnabled: true in settings.json not persisted across sessions — Voice Mode requires manual /voice on every startup
Preflight Checklist
- [x] I have searched existing issues and this hasn't been reported yet
- [x] This is a single bug report (please file separate reports for different bugs)
- [x] I am using the latest version of Claude Code
What's Wrong?
Bug Description
The voiceEnabled: true setting in ~/.claude/settings.json is not respected when
starting a new Claude Code session. Voice Mode is disabled on every startup and
must be re-enabled manually by running /voice each time.
Environment
| Field | Value |
|---|---|
| Claude Code version | 2.1.76 |
| OS | Windows 10 (MINGW64 / Git Bash — MINGW64_NT-10.0-26200) |
| Shell | bash (MSYS2 / Git Bash) |
| Node.js version | v22.19.0 |
## Configuration File
Path: ~/.claude/settings.json
```json
{
"permissions": {
"allow": ["mcp__pencil"]
},
"language": "Español",
"alwaysThinkingEnabled": false,
"effortLevel": "high",
"autoUpdatesChannel": "latest",
"voiceEnabled": true
}
### What Should Happen?
## Expected Behavior
If `"voiceEnabled": true` is present in `~/.claude/settings.json`, Voice Mode
should be active from the moment the session starts, without requiring the user to
run `/voice`.
## Actual Behavior
Voice Mode is disabled on every new session regardless of the `voiceEnabled`
setting. The user must run `/voice` manually on every startup.
### Steps to Reproduce
1. Set `"voiceEnabled": true` in `~/.claude/settings.json`
2. Start a new Claude Code session
3. Observe that Voice Mode is **not** active — the Space bar does not trigger
recording
4. Run `/voice` to enable it manually
5. Exit the session
6. Repeat from step 2 — Voice Mode is again disabled on the new session
### Claude Model
Sonnet (default)
### Is this a regression?
I don't know
### Last Working Version
_No response_
### Claude Code Version
2.1.76
### Platform
Anthropic API
### Operating System
Windows
### Terminal/Shell
VS Code integrated terminal
### Additional Information
Additional Context
- The setting voiceEnabled is present in the file and set to true before the
session starts.
- Running /voice within the session works correctly and enables Voice Mode for that
session only.
- The issue appears to be that the setting is either not being read at startup or
not being applied to initialize the Voice Mode state.
- Reproducible consistently across multiple sessions.
Workaround
Manually run /voice at the beginning of every session.This issue has 3 comments on GitHub. Read the full discussion on GitHub ↗