@domdhi/claude-code-tts 1.0.0

package/repeat.py

@@ -0,0 +1,108 @@
+#!/usr/bin/env python3
+"""
+claude-code-tts — UserPromptSubmit hook
+Intercepts /repeat and /voice:stop commands.
+  /repeat       → replays the last spoken response
+  /voice:stop   → stops speech immediately
+"""
+
+import json
+import sys
+import os
+import socket
+import subprocess
+import time
+
+HOOK_DIR = os.path.dirname(os.path.abspath(__file__))
+ON_FILE = os.path.join(HOOK_DIR, 'on')
+LAST_FILE = os.path.join(HOOK_DIR, 'last.txt')
+VOICES_FILE = os.path.join(HOOK_DIR, 'voices.json')
+DAEMON_SCRIPT = os.path.join(HOOK_DIR, 'daemon.py')
+
+DAEMON_HOST = '127.0.0.1'
+DAEMON_PORT = 6254
+
+
+def _start_daemon():
+    kwargs = {
+        'stdout': subprocess.DEVNULL,
+        'stderr': subprocess.DEVNULL,
+    }
+    if sys.platform == 'win32':
+        kwargs['creationflags'] = subprocess.CREATE_NO_WINDOW
+    subprocess.Popen([sys.executable, DAEMON_SCRIPT], **kwargs)
+    for _ in range(40):
+        time.sleep(0.2)
+        try:
+            s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+            s.settimeout(0.5)
+            s.connect((DAEMON_HOST, DAEMON_PORT))
+            s.close()
+            return True
+        except Exception:
+            pass
+    return False
+
+
+def send_to_daemon(cmd_dict):
+    for attempt in range(2):
+        try:
+            s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+            s.settimeout(3.0)
+            s.connect((DAEMON_HOST, DAEMON_PORT))
+            s.send(json.dumps(cmd_dict).encode() + b'\n')
+            resp = s.recv(1024)
+            s.close()
+            return json.loads(resp.decode().strip()).get('ok', False)
+        except ConnectionRefusedError:
+            if attempt == 0:
+                _start_daemon()
+        except Exception:
+            return False
+    return False
+
+
+def load_default_voice():
+    if os.path.exists(VOICES_FILE):
+        try:
+            with open(VOICES_FILE, 'r', encoding='utf-8') as f:
+                cfg = json.load(f).get('default', {})
+                return cfg.get('voice', 'af_heart'), float(cfg.get('speed', 1.0))
+        except Exception:
+            pass
+    return 'af_heart', 1.0
+
+
+def main():
+    try:
+        data = json.loads(sys.stdin.read())
+    except Exception:
+        sys.exit(0)
+
+    prompt = data.get('prompt', '').strip()
+
+    if prompt in ('/voice stop', '/voice:stop', '/stop'):
+        send_to_daemon({'cmd': 'stop'})
+        print(json.dumps({'decision': 'block', 'reason': ''}))
+        sys.exit(0)
+
+    if prompt not in ('/repeat', '/voice repeat', '/voice:repeat'):
+        sys.exit(0)
+
+    # Replay last spoken text
+    if os.path.exists(LAST_FILE):
+        try:
+            with open(LAST_FILE, 'r', encoding='utf-8') as f:
+                text = f.read().strip()
+            if text:
+                voice, speed = load_default_voice()
+                send_to_daemon({'cmd': 'speak', 'text': text, 'voice': voice, 'speed': speed})
+        except Exception:
+            pass
+
+    print(json.dumps({'decision': 'block', 'reason': ''}))
+    sys.exit(0)
+
+
+if __name__ == '__main__':
+    main()

package/stop.py

@@ -0,0 +1,244 @@
+#!/usr/bin/env python3
+"""
+claude-code-tts — Stop hook
+Fires when Claude finishes a response. Reads the last assistant message from
+the transcript and sends it to the TTS daemon for playback.
+Starts the daemon automatically if it's not running.
+"""
+
+import json
+import sys
+import re
+import os
+import socket
+import subprocess
+import time
+
+HOOK_DIR = os.path.dirname(os.path.abspath(__file__))
+ON_FILE = os.path.join(HOOK_DIR, 'on')
+LAST_FILE = os.path.join(HOOK_DIR, 'last.txt')
+VOICES_FILE = os.path.join(HOOK_DIR, 'voices.json')
+DAEMON_SCRIPT = os.path.join(HOOK_DIR, 'daemon.py')
+
+DAEMON_HOST = '127.0.0.1'
+DAEMON_PORT = 6254
+
+READ_ALL_CHAIN = False  # True = speak all assistant messages in chain, False = last only
+
+
+def strip_markdown(text):
+    text = re.sub(r'```[\w]*\n.*?```', '[code block]', text, flags=re.DOTALL)
+    text = re.sub(r'`[^`]+`', lambda m: m.group(0)[1:-1], text)
+    text = re.sub(r'^#{1,6}\s+', '', text, flags=re.MULTILINE)
+    text = re.sub(r'\*{1,3}([^*]+)\*{1,3}', r'\1', text)
+    text = re.sub(r'_{1,3}([^_]+)_{1,3}', r'\1', text)
+    text = re.sub(r'\[([^\]]+)\]\([^\)]+\)', r'\1', text)
+    text = re.sub(r'^\s*[-*+]\s+', '', text, flags=re.MULTILINE)
+    text = re.sub(r'^\s*\d+\.\s+', '', text, flags=re.MULTILINE)
+    text = re.sub(r'^[-*_]{3,}$', '', text, flags=re.MULTILINE)
+    text = re.sub(r'<[^>]+>', '', text)
+    text = re.sub(r'\n{3,}', '\n\n', text)
+    text = _sanitize_unicode(text)
+    return text.strip()
+
+
+def _sanitize_unicode(text):
+    """Replace unicode symbols that TTS engines can't phonemize with ASCII equivalents."""
+    replacements = {
+        '→': ', ', '←': ', ', '↑': ', ', '↓': ', ',
+        '⇒': ', ', '⇐': ', ', '⇔': ', ',
+        '—': ', ', '–': ', ', '…': '...',
+        '\u2019': "'", '\u2018': "'",  # smart quotes
+        '\u201c': '"', '\u201d': '"',
+        '•': '', '·': '',
+        '✓': 'yes', '✗': 'no', '✅': 'yes', '❌': 'no',
+        '🔴': '', '🟡': '', '🟢': '', '⭐': '',
+    }
+    for src, dst in replacements.items():
+        text = text.replace(src, dst)
+    # Strip remaining non-ASCII
+    text = re.sub(r'[^\x00-\x7F]+', ' ', text)
+    text = re.sub(r' {2,}', ' ', text)
+    return text
+
+
+def extract_last_assistant_message(transcript_path):
+    if not os.path.exists(transcript_path):
+        return None
+    entries = []
+    with open(transcript_path, 'r', encoding='utf-8', errors='replace') as f:
+        for line in f:
+            line = line.strip()
+            if not line:
+                continue
+            try:
+                entries.append(json.loads(line))
+            except json.JSONDecodeError:
+                continue
+
+    # Find index of last real user message (skip tool result entries)
+    last_user_idx = -1
+    for i, obj in enumerate(entries):
+        if obj.get('type') == 'user':
+            content = obj.get('message', {}).get('content', [])
+            if isinstance(content, list) and content and all(
+                isinstance(b, dict) and b.get('type') == 'tool_result'
+                for b in content
+            ):
+                continue
+            last_user_idx = i
+
+    texts = []
+    for obj in entries[last_user_idx + 1:]:
+        if obj.get('type') != 'assistant':
+            continue
+        content = obj.get('message', {}).get('content', '')
+        if isinstance(content, str) and content.strip():
+            texts.append(content.strip())
+        elif isinstance(content, list):
+            parts = [b.get('text', '') for b in content
+                     if isinstance(b, dict) and b.get('type') == 'text']
+            combined = '\n'.join(parts).strip()
+            if combined:
+                texts.append(combined)
+
+    if not texts:
+        return None
+    return '\n\n'.join(texts) if READ_ALL_CHAIN else texts[-1]
+
+
+def load_voices():
+    if os.path.exists(VOICES_FILE):
+        try:
+            with open(VOICES_FILE, 'r', encoding='utf-8') as f:
+                return json.load(f)
+        except Exception:
+            pass
+    return {'default': {'voice': 'af_heart', 'speed': 1.0}}
+
+
+def get_project_voice(transcript_path, voices):
+    """Look up voice by project dir encoded in transcript path.
+    Transcript paths look like: .../.claude/projects/c--Users-...-ProjectName/...
+    voices.json 'projects' keys are matched as substrings of that encoded segment."""
+    projects = voices.get('projects', {})
+    if not projects or not transcript_path:
+        return None
+    norm = transcript_path.replace('\\', '/').lower()
+    for key, cfg in projects.items():
+        if key.lower() in norm:
+            return cfg
+    return None
+
+
+def parse_agent_voice(text, voices, fallback_cfg=None):
+    """Parse [AgentName]: prefix. Returns (stripped_text, voice, speed).
+    Priority: agent prefix → fallback_cfg (project) → default."""
+    match = re.match(r'^\[([^\]]+)\]:\s*', text)
+    if match:
+        agent = match.group(1)
+        text = text[match.end():]
+        cfg = voices.get(agent) or fallback_cfg or voices.get('default') or {}
+    else:
+        cfg = fallback_cfg or voices.get('default') or {}
+    voice = cfg.get('voice', 'af_heart')
+    speed = float(cfg.get('speed', 1.0))
+    return text, voice, speed
+
+
+def _start_daemon():
+    """Start the TTS daemon as a detached background process."""
+    kwargs = {
+        'stdout': subprocess.DEVNULL,
+        'stderr': subprocess.DEVNULL,
+    }
+    if sys.platform == 'win32':
+        kwargs['creationflags'] = subprocess.CREATE_NO_WINDOW
+    subprocess.Popen([sys.executable, DAEMON_SCRIPT], **kwargs)
+    # Poll until daemon is ready (up to 8s for model load on first run)
+    for _ in range(40):
+        time.sleep(0.2)
+        try:
+            s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+            s.settimeout(0.5)
+            s.connect((DAEMON_HOST, DAEMON_PORT))
+            s.close()
+            return True
+        except Exception:
+            pass
+    return False
+
+
+def send_to_daemon(cmd_dict):
+    """Send a command to the daemon. Starts daemon if not running."""
+    for attempt in range(2):
+        try:
+            s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+            s.settimeout(3.0)
+            s.connect((DAEMON_HOST, DAEMON_PORT))
+            s.sendall(json.dumps(cmd_dict).encode() + b'\n')
+            resp = s.recv(1024)
+            s.close()
+            return json.loads(resp.decode().strip()).get('ok', False)
+        except ConnectionRefusedError:
+            if attempt == 0:
+                _start_daemon()
+        except Exception:
+            return False
+    return False
+
+
+def get_project_key(transcript_path):
+    """Extract encoded project dir from transcript path.
+    e.g. .../.claude/projects/c--Users-dbaca-Repos-MyProject/... → 'c--Users-dbaca-Repos-MyProject'"""
+    if not transcript_path:
+        return None
+    parts = transcript_path.replace('\\', '/').split('/')
+    try:
+        idx = parts.index('projects')
+        return parts[idx + 1]
+    except (ValueError, IndexError):
+        return None
+
+
+def speak(text, voice='af_heart', speed=1.0, project=None):
+    send_to_daemon({'cmd': 'speak', 'text': text, 'voice': voice, 'speed': speed, 'project': project})
+
+
+def main():
+    try:
+        data = json.loads(sys.stdin.read())
+    except Exception:
+        sys.exit(0)
+
+    if not os.path.exists(ON_FILE):
+        sys.exit(0)
+
+    transcript_path = data.get('transcript_path', '')
+    if not transcript_path:
+        sys.exit(0)
+
+    time.sleep(0.5)
+    text = extract_last_assistant_message(transcript_path)
+    if not text:
+        sys.exit(0)
+
+    cleaned = strip_markdown(text)
+    if not cleaned:
+        sys.exit(0)
+
+    try:
+        with open(LAST_FILE, 'w', encoding='utf-8') as f:
+            f.write(cleaned)
+    except Exception:
+        pass
+
+    voices = load_voices()
+    proj_cfg = get_project_voice(transcript_path, voices)
+    cleaned, voice, speed = parse_agent_voice(cleaned, voices, fallback_cfg=proj_cfg)
+    speak(cleaned, voice, speed, project=get_project_key(transcript_path))
+    sys.exit(0)
+
+
+if __name__ == '__main__':
+    main()

package/task-hook.py

@@ -0,0 +1,171 @@
+#!/usr/bin/env python3
+"""
+claude-code-tts — PostToolUse:Task hook
+Fires after a Task tool call completes. Reads subagent_type from tool_input
+to look up the agent's voice — no [agent-name]: prefix required.
+"""
+
+import json
+import sys
+import re
+import os
+import socket
+import subprocess
+import time
+
+HOOK_DIR = os.path.dirname(os.path.abspath(__file__))
+ON_FILE = os.path.join(HOOK_DIR, 'on')
+VOICES_FILE = os.path.join(HOOK_DIR, 'voices.json')
+DAEMON_SCRIPT = os.path.join(HOOK_DIR, 'daemon.py')
+
+DAEMON_HOST = '127.0.0.1'
+DAEMON_PORT = 6254
+
+
+def strip_markdown(text):
+    text = re.sub(r'```[\w]*\n.*?```', '[code block]', text, flags=re.DOTALL)
+    text = re.sub(r'`[^`]+`', lambda m: m.group(0)[1:-1], text)
+    text = re.sub(r'^#{1,6}\s+', '', text, flags=re.MULTILINE)
+    text = re.sub(r'\*{1,3}([^*]+)\*{1,3}', r'\1', text)
+    text = re.sub(r'_{1,3}([^_]+)_{1,3}', r'\1', text)
+    text = re.sub(r'\[([^\]]+)\]\([^\)]+\)', r'\1', text)
+    text = re.sub(r'^\s*[-*+]\s+', '', text, flags=re.MULTILINE)
+    text = re.sub(r'^\s*\d+\.\s+', '', text, flags=re.MULTILINE)
+    text = re.sub(r'^[-*_]{3,}$', '', text, flags=re.MULTILINE)
+    text = re.sub(r'<[^>]+>', '', text)
+    text = re.sub(r'\n{3,}', '\n\n', text)
+    text = _sanitize_unicode(text)
+    return text.strip()
+
+
+def _sanitize_unicode(text):
+    replacements = {
+        '→': ', ', '←': ', ', '↑': ', ', '↓': ', ',
+        '⇒': ', ', '⇐': ', ', '⇔': ', ',
+        '—': ', ', '–': ', ', '…': '...',
+        '\u2019': "'", '\u2018': "'",
+        '\u201c': '"', '\u201d': '"',
+        '•': '', '·': '',
+        '✓': 'yes', '✗': 'no', '✅': 'yes', '❌': 'no',
+        '🔴': '', '🟡': '', '🟢': '', '⭐': '',
+    }
+    for src, dst in replacements.items():
+        text = text.replace(src, dst)
+    text = re.sub(r'[^\x00-\x7F]+', ' ', text)
+    text = re.sub(r' {2,}', ' ', text)
+    return text
+
+
+def extract_text_from_response(tool_response):
+    """Pull plain text out of whatever shape the tool_response takes."""
+    if isinstance(tool_response, str):
+        return tool_response.strip()
+
+    if isinstance(tool_response, dict):
+        content = tool_response.get('content', tool_response.get('output', ''))
+        if isinstance(content, str):
+            return content.strip()
+        if isinstance(content, list):
+            parts = []
+            for block in content:
+                if isinstance(block, dict):
+                    parts.append(block.get('text', ''))
+                elif isinstance(block, str):
+                    parts.append(block)
+            return '\n'.join(p for p in parts if p).strip()
+
+    return ''
+
+
+def load_voices():
+    if os.path.exists(VOICES_FILE):
+        try:
+            with open(VOICES_FILE, 'r', encoding='utf-8') as f:
+                return json.load(f)
+        except Exception:
+            pass
+    return {'default': {'voice': 'af_heart', 'speed': 1.0}}
+
+
+def _start_daemon():
+    kwargs = {
+        'stdout': subprocess.DEVNULL,
+        'stderr': subprocess.DEVNULL,
+    }
+    if sys.platform == 'win32':
+        kwargs['creationflags'] = subprocess.CREATE_NO_WINDOW
+    subprocess.Popen([sys.executable, DAEMON_SCRIPT], **kwargs)
+    for _ in range(40):
+        time.sleep(0.2)
+        try:
+            s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+            s.settimeout(0.5)
+            s.connect((DAEMON_HOST, DAEMON_PORT))
+            s.close()
+            return True
+        except Exception:
+            pass
+    return False
+
+
+def send_to_daemon(cmd_dict):
+    for attempt in range(2):
+        try:
+            s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+            s.settimeout(3.0)
+            s.connect((DAEMON_HOST, DAEMON_PORT))
+            s.sendall(json.dumps(cmd_dict).encode() + b'\n')
+            resp = s.recv(1024)
+            s.close()
+            return json.loads(resp.decode().strip()).get('ok', False)
+        except ConnectionRefusedError:
+            if attempt == 0:
+                _start_daemon()
+        except Exception:
+            return False
+    return False
+
+
+def main():
+    try:
+        data = json.loads(sys.stdin.read())
+    except Exception:
+        sys.exit(0)
+
+    if not os.path.exists(ON_FILE):
+        sys.exit(0)
+
+    if data.get('tool_name') != 'Task':
+        sys.exit(0)
+
+    # Determine agent from tool_input.subagent_type — no prefix required
+    tool_input = data.get('tool_input', {})
+    agent_name = tool_input.get('subagent_type', '')
+
+    tool_response = data.get('tool_response', '')
+    raw_text = extract_text_from_response(tool_response)
+
+    if not raw_text:
+        sys.exit(0)
+
+    # Strip [agent-name]: prefix if the agent included it anyway
+    raw_text = re.sub(r'^\[[^\]]+\]:\s*', '', raw_text)
+
+    cleaned = strip_markdown(raw_text)
+    if not cleaned:
+        sys.exit(0)
+
+    voices = load_voices()
+    cfg = voices.get(agent_name) or voices.get('default') or {}
+    voice = cfg.get('voice', 'af_heart')
+    speed = float(cfg.get('speed', 1.0))
+
+    # Use tool_use_id as a stable per-task identity so rapid task completions
+    # replace each other in the queue rather than stacking.
+    task_id = data.get('tool_use_id') or data.get('session_id') or 'task'
+    send_to_daemon({'cmd': 'speak', 'text': cleaned, 'voice': voice, 'speed': speed, 'project': task_id})
+    sys.exit(0)
+
+
+if __name__ == '__main__':
+    main()

package/voices.json

@@ -0,0 +1,14 @@
+{
+  "default": {
+    "voice": "af_heart",
+    "speed": 1.0
+  },
+  "_comment_agents": "Add agent name keys to route agent voices. Key = subagent_type value (e.g. 'general-purpose') or [AgentName]: prefix in response (lowercase).",
+  "_comment_projects": "Add a 'projects' section to route by project. Key = substring of ~/.claude/projects/ dir name (case-insensitive).",
+  "_examples": {
+    "general-purpose": {"voice": "am_michael", "speed": 1.0},
+    "projects": {
+      "my-project": {"voice": "am_onyx", "speed": 0.95}
+    }
+  }
+}