136 lines
5.6 KiB
Python
136 lines
5.6 KiB
Python
import logging
|
|
from typing import Dict, Any
|
|
from google.cloud import texttospeech
|
|
|
|
from core.speech_service import TextToSpeechService, BaseAIConversationService, BaseConversationFlowService
|
|
from .models import SCENARIO_PERSONALITIES, Personality
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
class IndonesianTextToSpeechService(TextToSpeechService):
|
|
def __init__(self):
|
|
super().__init__(language_code="id-ID")
|
|
|
|
def _get_voice_config(self, gender: str, character_name: str = None) -> Dict[str, Any]:
|
|
"""Get Indonesian-specific voice configuration."""
|
|
tts_gender = self.gender_map.get(gender, texttospeech.SsmlVoiceGender.FEMALE)
|
|
|
|
character_voice_map = {
|
|
"Pak Budi": {
|
|
"name": "id-ID-Chirp3-HD-Charon",
|
|
"speaking_rate": 0.95,
|
|
"pitch": None,
|
|
"ssml_gender": texttospeech.SsmlVoiceGender.MALE,
|
|
},
|
|
"Ibu Sari": {
|
|
"name": "id-ID-Chirp3-HD-Kore",
|
|
"speaking_rate": 1.0,
|
|
"pitch": None,
|
|
"ssml_gender": texttospeech.SsmlVoiceGender.FEMALE,
|
|
},
|
|
"Mbak Sari": {
|
|
"name": "id-ID-Chirp3-HD-Zephyr",
|
|
"speaking_rate": 1.1,
|
|
"pitch": None,
|
|
"ssml_gender": texttospeech.SsmlVoiceGender.FEMALE,
|
|
},
|
|
"Adik Kasir": {
|
|
"name": "id-ID-Chirp3-HD-Aoede",
|
|
"speaking_rate": 1.05,
|
|
"pitch": None,
|
|
"ssml_gender": texttospeech.SsmlVoiceGender.FEMALE,
|
|
},
|
|
"Tetangga Ali": {
|
|
"name": "id-ID-Chirp3-HD-Puck",
|
|
"speaking_rate": 1.05,
|
|
"pitch": None,
|
|
"ssml_gender": texttospeech.SsmlVoiceGender.MALE,
|
|
}
|
|
}
|
|
|
|
gender_voice_fallback = {
|
|
texttospeech.SsmlVoiceGender.MALE: {
|
|
"name": "id-ID-Chirp3-HD-Fenrir",
|
|
"speaking_rate": 1.0,
|
|
"pitch": None,
|
|
"ssml_gender": texttospeech.SsmlVoiceGender.MALE,
|
|
},
|
|
texttospeech.SsmlVoiceGender.FEMALE: {
|
|
"name": "id-ID-Chirp3-HD-Leda",
|
|
"speaking_rate": 1.0,
|
|
"pitch": None,
|
|
"ssml_gender": texttospeech.SsmlVoiceGender.FEMALE,
|
|
}
|
|
}
|
|
|
|
if character_name and character_name in character_voice_map:
|
|
config_set = character_voice_map[character_name]
|
|
logger.info(f"Using character-specific voice for '{character_name}': {config_set['name']}")
|
|
return config_set
|
|
|
|
config_set = gender_voice_fallback.get(tts_gender, gender_voice_fallback[texttospeech.SsmlVoiceGender.FEMALE])
|
|
logger.info(f"Using gender fallback voice for {tts_gender}: {config_set['name']}")
|
|
return config_set
|
|
|
|
class IndonesianAIConversationService(BaseAIConversationService):
|
|
def __init__(self):
|
|
super().__init__(language_code="id")
|
|
|
|
def get_personality_for_scenario(self, scenario: str, character_name: str = None) -> Personality:
|
|
"""Get Indonesian personality based on scenario and character name."""
|
|
if scenario in SCENARIO_PERSONALITIES:
|
|
personalities = SCENARIO_PERSONALITIES[scenario]
|
|
if character_name and character_name in personalities:
|
|
return personalities[character_name]
|
|
else:
|
|
return list(personalities.values())[0]
|
|
|
|
# Return default personality if scenario not found
|
|
from .models import Personality, CharacterType, Gender, PersonalityTone, GoalItem, HelpfulPhrase
|
|
return Personality(
|
|
character_type=CharacterType.GENERIC,
|
|
name="Pak/Bu",
|
|
gender=Gender.FEMALE,
|
|
tone=PersonalityTone.FRIENDLY,
|
|
age_range="middle-aged",
|
|
background="Helpful Indonesian person",
|
|
typical_phrases=["Halo!", "Apa kabar?", "Bisa saya bantu?"],
|
|
response_style="Friendly and helpful",
|
|
location_context="Indonesia",
|
|
scenario_title="General Conversation",
|
|
scenario_description="General Indonesian conversation practice",
|
|
scenario_challenge="Practice basic Indonesian conversation",
|
|
scenario_goal="Have a natural conversation",
|
|
goal_items=[],
|
|
helpful_phrases=[],
|
|
language_code="id",
|
|
country_code="Indonesia",
|
|
is_helpful=True,
|
|
is_talkative=True
|
|
)
|
|
|
|
class IndonesianConversationFlowService(BaseConversationFlowService):
|
|
def __init__(self):
|
|
super().__init__(language_code="id-ID")
|
|
self.tts_service = IndonesianTextToSpeechService()
|
|
self.ai_service = IndonesianAIConversationService()
|
|
|
|
def extract_scenario_from_context(self, context: str) -> str:
|
|
"""Extract scenario type from context string."""
|
|
logger.info(f"Extracting scenario from context: '{context}'")
|
|
context_lower = context.lower()
|
|
|
|
detected_scenario = None
|
|
if "coffee_shop" in context_lower or "coffee" in context_lower:
|
|
detected_scenario = "coffee_shop"
|
|
elif "warung" in context_lower or "nasi goreng" in context_lower:
|
|
detected_scenario = "warung"
|
|
elif "ojek" in context_lower or "mall" in context_lower:
|
|
detected_scenario = "ojek"
|
|
elif "alfamart" in context_lower or "indomie" in context_lower:
|
|
detected_scenario = "alfamart"
|
|
else:
|
|
detected_scenario = "warung" # Default to warung
|
|
|
|
logger.info(f"Detected scenario: '{detected_scenario}'")
|
|
return detected_scenario |