@domdhi/claude-code-tts 1.4.0 → 2.0.1

package/.claude/commands/voice/off.md

@@ -1,7 +1,28 @@
-Disable TTS voice output by removing the enable file. Run the appropriate command for your OS silently using the Bash tool, then respond with only: "TTS disabled."
+Disable TTS voice output. Run the following Python command using the Bash tool (it stops any playing audio and removes the enable file), then respond with only: "TTS disabled."
 
-Mac/Linux:
-rm -f ~/.claude/hooks/tts/on
+```bash
+python -c "
+import socket, json, hashlib, os
 
-Windows:
-del /F "%USERPROFILE%\.claude\hooks\tts\on" 2>nul
+for hook_dir in [
+    os.path.join(os.path.expanduser('~'), '.claude', 'hooks', 'tts'),
+    os.path.join(os.getcwd(), '.claude', 'hooks', 'tts'),
+]:
+    if not os.path.isdir(hook_dir):
+        continue
+    port = 49152 + (int(hashlib.md5(hook_dir.encode()).hexdigest(), 16) % 16384)
+    try:
+        s = socket.socket()
+        s.settimeout(1)
+        s.connect(('127.0.0.1', port))
+        s.sendall(json.dumps({'cmd': 'stop'}).encode() + b'\n')
+        s.close()
+    except Exception:
+        pass
+    on_file = os.path.join(hook_dir, 'on')
+    try:
+        os.remove(on_file)
+    except Exception:
+        pass
+" 2>/dev/null || true
+```

package/INSTALL.md

@@ -36,11 +36,11 @@ Requires Node.js 16+ and Python 3.10+. No cloning required.
 The installer automatically:
 1. Checks Python version (3.10+ required)
 2. Installs required packages (`edge-tts`, `miniaudio`, `sounddevice`, `cffi`)
-3. Copies hook scripts to `~/.claude/hooks/tts/`
+3. Copies hook scripts to the install directory (`.claude/hooks/tts/` locally, `~/.claude/hooks/tts/` globally)
 4. Creates the `on` file (TTS enabled immediately)
 5. Optionally installs kokoro-onnx offline fallback (~82MB)
 6. Installs `/voice:*` slash commands
-7. Patches `settings.json` / `settings.local.json` with hook entries (backs up original first)
+7. Patches `settings.local.json` / `settings.json` with hook entries (backs up original first)
 
 **Local vs global:**
 - Default (no flag): everything goes into `./.claude/` — hook scripts, commands, and `settings.local.json` (gitignored). Good for shipping TTS config alongside a project. Each install gets its own daemon port, so multiple local-installed projects run independently without conflict.
@@ -356,6 +356,11 @@ The daemon auto-restarts on the next Claude response.
 - kokoro-onnx has a 510-token (~1500 char) hard limit
 - The daemon chunks text at sentence boundaries automatically — if you're hitting this, check `debug.log` for `IndexError`
 
+### Windows: hook command not found / `C:Python313python.exe`
+- Claude Code runs hooks via bash, which cannot resolve `C:\...\python.exe` as a command
+- Hook commands must use `python` (on PATH), not the full Windows executable path
+- The installer handles this automatically; for manual edits use `python "C:\...\script.py"` not `C:\Python313\python.exe "C:\...\script.py"`
+
 ### Windows: DETACHED_PROCESS causes silence
 - Do not add `DETACHED_PROCESS` to the subprocess flags — it breaks the Windows audio session
 - `CREATE_NO_WINDOW` only is correct (already set in the hook files)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@domdhi/claude-code-tts",
-  "version": "1.4.0",
+  "version": "2.0.1",
   "description": "Neural TTS hook system for Claude Code. Reads Claude's responses aloud as they finish.",
   "bin": {
     "claude-code-tts": "bin/install.js"