#!/usr/bin/env python3 """ CALMHIVE Enhanced Trigger Words for Voice System Created: May 15, 2025 This module provides enhanced trigger word detection for the CALMHIVE voice system. It includes a comprehensive list of trigger words and variations to improve recognition. """ import re # Define variations of trigger words to enhance recognition TRIGGER_VARIATIONS = { # Primary assistant name variations "claude": ["claude", "cloud", "clod", "clawed", "cloud", "clout", "clown", "claud", "clod", "clawed", "claude's", "clouds", "cloudy"], # Friendly assistant variations "friend": ["friend", "friends", "friendly", "befriend", "my friend", "hey friend", "fiend", "fren", "front", "trend", "end", "rent", "fred", "fried"], # Code assistant variations "code": ["code", "coder", "coding", "encode", "decode", "coded", "codes", "code helper", "code buddy", "let's code", "kode", "coat", "cold"], # CALMHIVE specific variations "calmhive": ["calmhive", "calm hive", "calm-hive", "call hive", "call mine", "calm mind", "calm five", "com hive", "bee hive", "bee helper"], # Generic assistant variations "assistant": ["assistant", "assist", "assistance", "assistive", "assisting", "assists", "assist me", "assistant please", "my assistant", "hey assistant"], # Action-based triggers "help": ["help", "helper", "helping", "helped", "helps", "help me", "can you help", "please help", "i need help", "would you help"], # Common voice assistant activations "hey": ["hey", "hi", "hello", "hey there", "hey you", "hey now", "hay", "they", "hay"], # OK prefix variations (like "OK Google") "ok": ["ok", "okay", "ok now", "alright", "all right", "o.k.", "ok then"], # Voice specific activations "voice": ["voice", "voice assistant", "voice helper", "voice mode", "by voice", "with voice", "through voice", "using voice"], # Generic command triggers "command": ["command", "do this", "execute", "run", "perform", "please", "can you", "would you", "could you", "execute command", "run command"], } # Additional patterns for compound triggers COMPOUND_PATTERNS = [ # Greeting + trigger combinations (r"(hey|hi|hello|ok|okay)\s+(claude|friend|assistant|calmhive|code)", 0.9), # Action phrases (r"(let's|let us)\s+(code|build|create|make|generate)", 0.8), (r"(help|assist)\s+(me|us)\s+with", 0.8), (r"(can|could|would)\s+you\s+(help|assist|show|tell|explain)", 0.7), # Questions directed at the assistant (r"(what|how|why|when|where|who)\s+(can|should|would|could|do|does|is|are)\s+", 0.6), # Polite requests (r"(please|kindly|can you please|would you please)\s+(help|show|tell|explain|create)", 0.7), # Direct commands (r"^(find|create|build|make|show|tell|explain|write)\s+", 0.6), ] # Flatten the trigger variations into a single list ALL_TRIGGERS = [] for primary, variations in TRIGGER_VARIATIONS.items(): ALL_TRIGGERS.append(primary) ALL_TRIGGERS.extend(variations) # Remove duplicates while preserving order ALL_TRIGGERS = list(dict.fromkeys(ALL_TRIGGERS)) def contains_trigger_word(text, threshold=0.6): """ Enhanced function to check if text contains any trigger word or variation. Uses a more sophisticated approach with confidence scoring: 1. Direct matches of trigger words 2. Compound pattern matching with confidence weights 3. Partial matches for similar sounding words (handles speech recognition errors) Args: text: The text to check for trigger words threshold: Confidence threshold (0.0 to 1.0) for considering a match Returns: Boolean indicating if a trigger was detected with sufficient confidence """ if not text: return False # Convert text to lowercase for case-insensitive matching text_lower = text.lower() # Track confidence score for potential triggers max_confidence = 0.0 # Simple greeting words should not trigger by themselves if text_lower.strip() in ["hello", "hi", "hey", "ok", "okay"]: return False # 1. Check for exact trigger matches (highest confidence) for primary, variations in TRIGGER_VARIATIONS.items(): # Check primary trigger (highest confidence) if f" {primary} " in f" {text_lower} ": return True # Primary trigger with perfect match # Check for exact matches of variations (high confidence) for variant in variations: if variant in text_lower: # Check if it's a standalone word if re.search(r'\b' + re.escape(variant) + r'\b', text_lower): max_confidence = max(max_confidence, 0.9) else: # Partial match (embedded in other words) max_confidence = max(max_confidence, 0.7) # 2. Check compound patterns for pattern, confidence in COMPOUND_PATTERNS: if re.search(pattern, text_lower): max_confidence = max(max_confidence, confidence) # 3. Check for partial matches with word similarity words = text_lower.split() # Longer phrases increase confidence in trigger detection if len(words) >= 3: # Check each word against triggers for word in words: # Skip very short words if len(word) <= 2: continue for trigger in ALL_TRIGGERS: # Skip very short triggers if len(trigger) <= 2: continue # Check for partial matches at word beginnings # E.g. "clau" could be a partial "claude" if len(word) >= 3 and len(trigger) >= 3: min_len = min(len(word), len(trigger)) if word[:min_len] == trigger[:min_len]: # Calculate match percentage similarity = (min_len) / max(len(word), len(trigger)) match_confidence = similarity * 0.8 # Scale confidence max_confidence = max(max_confidence, match_confidence) # Return True if confidence exceeds threshold return max_confidence >= threshold # Test function def test_trigger_detection(): """Test function for trigger word detection""" test_cases = [ ("Hello Claude, how are you today?", True), ("Hey friend, can you help me with this?", True), ("Let's code a new website", True), ("I need help with my homework", False), # No clear trigger ("Hey Clod, are you there?", True), # Misspelled but similar ("Just saying hi", False), ("Calmhive, what's the weather?", True), ("Call hive for assistance", True), # Similar sounding ("Ok computer", False), # Wrong assistant name ("Hey there Claude", True), ] success = 0 for text, expected in test_cases: result = contains_trigger_word(text) if result == expected: success += 1 status = "✓" else: status = "✗" print(f"{status} Text: \"{text}\" -> Detected: {result}, Expected: {expected}") print(f"Passed {success}/{len(test_cases)} tests") if __name__ == "__main__": # Run test function test_trigger_detection()