Merge pull request #78 from rimgosu/main

feat: naver voice

app/api/api_v1/endpoints/gree.py

@@ -16,6 +16,7 @@
 import uuid
 
 from app.schemas.greeFileDto import GreeFileSchema
+from app.services.gree_update_service import update_gree_voice_type
 from app.services.image_service import create_image, check_image_status, upload_images_to_azure
 from app.services.upload_service import upload_file_to_azure, upload_greefile_to_azure, \
     upload_yaml_to_azure_blob, upload_gif_to_azure_blob
@@ -107,6 +108,8 @@ async def update_gree(gree_id: int, gree_update: GreeUpdate, current_user: Membe
     if not gree:
         raise HTTPException(status_code=404, detail="Gree not found")
 
+    await update_gree_voice_type(db, gree_id, gree_update)
+
     updated_gree = await crud_update_gree(db, gree_id, gree_update)
     return {"message": "Gree updated successfully"}
 

app/models/enums.py

@@ -41,9 +41,25 @@ class EmotionTypeEnum(PyEnum):
 
 @unique
 class VoiceTypeEnum(PyEnum):
-    ALLOY = 'alloy'
-    ECHO = 'echo'
-    FABLE = 'fable'
-    ONYX = 'onyx'
-    NOVA = 'nova'
-    SHIMMER = 'shimmer'
+    NWOOF = 'nwoof'
+    NJONGHYUN = 'njonghyun'
+    VDONGHYUN = 'vdonghyun'
+    NMAMMON = 'nmammon'
+    NGARAM = 'ngaram'
+    NMEOW = 'nmeow'
+    NIHYUN = 'nihyun'
+    VYUNA = 'vyuna'
+    VHYERI = 'vhyeri'
+    NHAJUN = 'nhajun'
+    NJAEWOOK = 'njaewook'
+    NJOONYOUNG = 'njoonyoung'
+    VDAESEONG = 'vdaeseong'
+    VIAN = 'vian'
+    NKYUWON = 'nkyuwon'
+    VDAIN = 'vdain'
+    NDAIN = 'ndain'
+    NMINSEO = 'nminseo'
+    NBORA = 'nbora'
+    NJIWON = 'njiwon'
+    VGOEUN = 'vgoeun'
+    NTIFFANY = 'ntiffany'

app/models/models.py

@@ -35,7 +35,7 @@ class Gree(Base):
     status = Column(Enum(StatusEnum))
     isFavorite = Column(Boolean, default=False)
     # 그리는 프롬프트 엔지니어링을 통해 다음과 같이 TTS의 목소리가 결정되어야한다.
-    voice_type = Column(Enum(VoiceTypeEnum), default=VoiceTypeEnum.ALLOY)
+    voice_type = Column(Enum(VoiceTypeEnum), default=VoiceTypeEnum.NWOOF)
     register_at = Column(DateTime, nullable=False, default=datetime.now())
 
     member = relationship("Member", back_populates="gree")

app/services/gree_update_service.py

@@ -0,0 +1,53 @@
+from sqlalchemy import update
+from sqlalchemy.future import select
+from sqlalchemy.ext.asyncio import AsyncSession
+from app.models.models import Gree
+from app.schemas.greeDto import GreeUpdate
+import random  # 랜덤 모듈 import
+
+async def update_gree_voice_type(db: AsyncSession, gree_id: int, gree_update: GreeUpdate):
+    # MBTI의 첫 글자(E/I)에 따라 활기찬(E) 또는 차분한(I) 결정
+    personality_type = 'E' if gree_update.prompt_mbti and gree_update.prompt_mbti.startswith('E') else 'I'
+
+    # 성별에 따른 분류
+    gender = gree_update.prompt_gender.upper() if gree_update.prompt_gender else 'N'  # Default to 'N' if None
+
+    # 나이에 따른 분류
+    if gree_update.prompt_age is not None:
+        if gree_update.prompt_age <= 11:
+            age_group = '어린이'
+        elif 11 < gree_update.prompt_age <= 17:
+            age_group = '청소년'
+        else:
+            age_group = '청년'
+    else:
+        age_group = None
+
+    # VoiceTypeEnum에서 적절한 값을 찾기
+    voice_type = None
+    # 나이 그룹별로 분류된 VoiceTypeEnum 매핑
+    voice_type_mapping = {
+        ('E', '남자', '어린이'): ['nwoof'],
+        ('E', '남자', '청소년'): ['njonghyun'],
+        ('E', '남자', '청년'): ['vdonghyun', 'nmammon'],
+        ('E', '여자', '어린이'): ['ngaram', 'nmeow'],
+        ('E', '여자', '청소년'): ['nihyun'],
+        ('E', '여자', '청년'): ['vyuna', 'vhyeri'],
+        ('I', '남자', '어린이'): ['nhajun'],
+        ('I', '남자', '청소년'): ['njaewook', 'njoonyoung'],
+        ('I', '남자', '청년'): ['vdaeseong', 'vian', 'nkyuwon'],
+        ('I', '여자', '어린이'): ['vdain', 'ndain'],
+        ('I', '여자', '청소년'): ['nminseo', 'nbora', 'njiwon'],
+        ('I', '여자', '청년'): ['vgoeun', 'ntiffany'],
+    }
+
+    if age_group:
+        voice_options = voice_type_mapping.get((personality_type, gender, age_group), [])
+        if voice_options:
+            voice_type = random.choice(voice_options)  # 선택지 중 하나를 랜덤으로 선택
+
+    if voice_type:
+        # gree 객체의 voice_type 업데이트
+        stmt = update(Gree).where(Gree.id == gree_id).values(voice_type=voice_type)
+        await db.execute(stmt)
+        await db.commit()

app/services/voice_service.py

@@ -1,3 +1,4 @@
+import urllib.request
 import os
 import uuid
 import aiohttp
@@ -97,7 +98,9 @@ async def chat_with_openai_service(db: AsyncSession, chat_request: ChatRequestDt
         )
         gree_talk= completion.choices[0].message.content
 
-        mp3_path = await text_to_speech(gree.voice_type, gree_talk)
+        print(f'gree_talk = {gree_talk}')
+
+        mp3_path = await text_to_speech_naver(gree.voice_type, gree_talk)
         voice_url_azure= await upload_mp3_azure(mp3_path)
 
         createGptLogDto = CreateGreeTalkLogDto(
@@ -147,7 +150,39 @@ async def text_to_speech(voice_type: VoiceTypeEnum, gree_talk: str) -> str:
                 return mp3_path
             else:
                 return "Error: 응답 상태가 200이 아닙니다."
-
+
+async def text_to_speech_naver(voice_type: VoiceTypeEnum, text: str) -> str:
+    client_id = os.getenv('NAVER_CLIENT_ID')  # 환경변수에서 클라이언트 ID를 가져옵니다.
+    client_secret = os.getenv('NAVER_CLIENT_SECRET')  # 환경변수에서 클라이언트 비밀을 가져옵니다.
+    encText = urllib.parse.quote(text)
+    data = f"speaker={voice_type.value}&volume=0&speed=0&pitch=0&format=mp3&text=" + encText
+    url = "https://naveropenapi.apigw.ntruss.com/tts-premium/v1/tts"
+
+
+    headers = {
+        "X-NCP-APIGW-API-KEY-ID": client_id,
+        "X-NCP-APIGW-API-KEY": client_secret,
+        "Content-Type": "application/x-www-form-urlencoded"
+    }
+
+    async with aiohttp.ClientSession() as session:
+        async with session.post(url, data=data.encode('utf-8'), headers=headers) as response:
+            if response.status == 200:
+                content = await response.read()
+
+                # 지정된 경로에 디렉토리가 없으면 생성
+                dir_path = '/temp/voice_temp/'
+                if not os.path.exists(dir_path):
+                    os.makedirs(dir_path)
+
+                # 파일 저장 경로 변경
+                mp3_path = os.path.join(dir_path, f'{voice_type.value}_speech.mp3')
+
+                with open(mp3_path, 'wb') as f:
+                    f.write(content)
+                return mp3_path
+            else:
+                return "Error: 응답 상태가 200이 아닙니다."
 
 async def upload_mp3_azure(mp3_path: str) -> str:
     try: