malek-messaoudii
commited on
Commit
·
56dc677
1
Parent(s):
674469e
Refactor audio processing and chatbot services; enhance STT and TTS functionalities with base64 support and session management
Browse files- config.py +2 -2
- main.py +29 -15
- models/audio.py +25 -39
- requirements.txt +5 -5
- routes/audio.py +70 -154
- services/chatbot_service.py +90 -60
- services/stt_service.py +83 -55
- services/tts_service.py +109 -37
config.py
CHANGED
|
@@ -17,8 +17,8 @@ PROJECT_ROOT = API_DIR.parent
|
|
| 17 |
|
| 18 |
# ============ HUGGING FACE ============
|
| 19 |
HUGGINGFACE_API_KEY = os.getenv("HUGGINGFACE_API_KEY", "")
|
| 20 |
-
HUGGINGFACE_STANCE_MODEL_ID = os.getenv("HUGGINGFACE_STANCE_MODEL_ID"
|
| 21 |
-
HUGGINGFACE_LABEL_MODEL_ID = os.getenv("HUGGINGFACE_LABEL_MODEL_ID"
|
| 22 |
|
| 23 |
# ============ API CONFIGURATION ============
|
| 24 |
API_TITLE = "NLP Debater - Voice Chatbot API"
|
|
|
|
| 17 |
|
| 18 |
# ============ HUGGING FACE ============
|
| 19 |
HUGGINGFACE_API_KEY = os.getenv("HUGGINGFACE_API_KEY", "")
|
| 20 |
+
HUGGINGFACE_STANCE_MODEL_ID = os.getenv("HUGGINGFACE_STANCE_MODEL_ID")
|
| 21 |
+
HUGGINGFACE_LABEL_MODEL_ID = os.getenv("HUGGINGFACE_LABEL_MODEL_ID")
|
| 22 |
|
| 23 |
# ============ API CONFIGURATION ============
|
| 24 |
API_TITLE = "NLP Debater - Voice Chatbot API"
|
main.py
CHANGED
|
@@ -58,7 +58,7 @@ from config import (
|
|
| 58 |
HOST, PORT, RELOAD,
|
| 59 |
CORS_ORIGINS, CORS_CREDENTIALS, CORS_METHODS, CORS_HEADERS,
|
| 60 |
PRELOAD_MODELS_ON_STARTUP, LOAD_STANCE_MODEL, LOAD_KPA_MODEL,
|
| 61 |
-
LOAD_STT_MODEL, LOAD_CHATBOT_MODEL
|
| 62 |
)
|
| 63 |
|
| 64 |
@asynccontextmanager
|
|
@@ -92,9 +92,10 @@ async def lifespan(app: FastAPI):
|
|
| 92 |
# Load STT Model (Speech-to-Text)
|
| 93 |
if LOAD_STT_MODEL:
|
| 94 |
try:
|
| 95 |
-
logger.info("Loading STT Model
|
| 96 |
-
from services.stt_service import
|
| 97 |
-
|
|
|
|
| 98 |
logger.info("✓ STT model loaded successfully")
|
| 99 |
except Exception as e:
|
| 100 |
logger.error(f"✗ STT model loading failed: {str(e)}")
|
|
@@ -102,9 +103,10 @@ async def lifespan(app: FastAPI):
|
|
| 102 |
# Load Chatbot Model
|
| 103 |
if LOAD_CHATBOT_MODEL:
|
| 104 |
try:
|
| 105 |
-
logger.info("Loading Chatbot Model
|
| 106 |
-
from services.chatbot_service import
|
| 107 |
-
|
|
|
|
| 108 |
logger.info("✓ Chatbot model loaded successfully")
|
| 109 |
except Exception as e:
|
| 110 |
logger.error(f"✗ Chatbot model loading failed: {str(e)}")
|
|
@@ -139,9 +141,16 @@ app.add_middleware(
|
|
| 139 |
)
|
| 140 |
|
| 141 |
# Include routers
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 142 |
try:
|
| 143 |
from routes.audio import router as audio_router
|
| 144 |
-
app.include_router(audio_router, prefix="/audio", tags=["Audio
|
| 145 |
logger.info("✓ Audio routes registered")
|
| 146 |
except Exception as e:
|
| 147 |
logger.warning(f"⚠️ Audio routes failed to load: {e}")
|
|
@@ -163,7 +172,8 @@ async def root():
|
|
| 163 |
"version": API_VERSION,
|
| 164 |
"docs": "/docs",
|
| 165 |
"endpoints": {
|
| 166 |
-
"
|
|
|
|
| 167 |
"health": "/health",
|
| 168 |
"models-status": "/models-status"
|
| 169 |
}
|
|
@@ -180,18 +190,22 @@ async def models_status():
|
|
| 180 |
status = {
|
| 181 |
"stt_model": "unknown",
|
| 182 |
"tts_engine": "gtts (free)",
|
| 183 |
-
"chatbot_model": "unknown"
|
|
|
|
|
|
|
| 184 |
}
|
| 185 |
|
| 186 |
try:
|
| 187 |
-
from services.stt_service import
|
| 188 |
-
|
|
|
|
| 189 |
except:
|
| 190 |
status["stt_model"] = "error"
|
| 191 |
|
| 192 |
try:
|
| 193 |
-
from services.chatbot_service import
|
| 194 |
-
|
|
|
|
| 195 |
except:
|
| 196 |
status["chatbot_model"] = "error"
|
| 197 |
|
|
@@ -220,4 +234,4 @@ if __name__ == "__main__":
|
|
| 220 |
port=PORT,
|
| 221 |
reload=RELOAD,
|
| 222 |
log_level="info"
|
| 223 |
-
)
|
|
|
|
| 58 |
HOST, PORT, RELOAD,
|
| 59 |
CORS_ORIGINS, CORS_CREDENTIALS, CORS_METHODS, CORS_HEADERS,
|
| 60 |
PRELOAD_MODELS_ON_STARTUP, LOAD_STANCE_MODEL, LOAD_KPA_MODEL,
|
| 61 |
+
LOAD_STT_MODEL, LOAD_CHATBOT_MODEL, STT_MODEL_ID, CHATBOT_MODEL_ID
|
| 62 |
)
|
| 63 |
|
| 64 |
@asynccontextmanager
|
|
|
|
| 92 |
# Load STT Model (Speech-to-Text)
|
| 93 |
if LOAD_STT_MODEL:
|
| 94 |
try:
|
| 95 |
+
logger.info(f"Loading STT Model: {STT_MODEL_ID}")
|
| 96 |
+
from services.stt_service import STTService
|
| 97 |
+
stt_service = STTService()
|
| 98 |
+
await stt_service.initialize()
|
| 99 |
logger.info("✓ STT model loaded successfully")
|
| 100 |
except Exception as e:
|
| 101 |
logger.error(f"✗ STT model loading failed: {str(e)}")
|
|
|
|
| 103 |
# Load Chatbot Model
|
| 104 |
if LOAD_CHATBOT_MODEL:
|
| 105 |
try:
|
| 106 |
+
logger.info(f"Loading Chatbot Model: {CHATBOT_MODEL_ID}")
|
| 107 |
+
from services.chatbot_service import ChatbotService
|
| 108 |
+
chatbot_service = ChatbotService()
|
| 109 |
+
await chatbot_service.initialize()
|
| 110 |
logger.info("✓ Chatbot model loaded successfully")
|
| 111 |
except Exception as e:
|
| 112 |
logger.error(f"✗ Chatbot model loading failed: {str(e)}")
|
|
|
|
| 141 |
)
|
| 142 |
|
| 143 |
# Include routers
|
| 144 |
+
try:
|
| 145 |
+
from routes.audio import router as chatbot_router
|
| 146 |
+
app.include_router(chatbot_router, prefix="/api/v1", tags=["Voice Chatbot"])
|
| 147 |
+
logger.info("✓ Chatbot routes registered")
|
| 148 |
+
except Exception as e:
|
| 149 |
+
logger.warning(f"⚠️ Chatbot routes failed to load: {e}")
|
| 150 |
+
|
| 151 |
try:
|
| 152 |
from routes.audio import router as audio_router
|
| 153 |
+
app.include_router(audio_router, prefix="/audio", tags=["Audio Processing"])
|
| 154 |
logger.info("✓ Audio routes registered")
|
| 155 |
except Exception as e:
|
| 156 |
logger.warning(f"⚠️ Audio routes failed to load: {e}")
|
|
|
|
| 172 |
"version": API_VERSION,
|
| 173 |
"docs": "/docs",
|
| 174 |
"endpoints": {
|
| 175 |
+
"voice_chatbot": "/api/v1/chat/message",
|
| 176 |
+
"audio_processing": "/docs#/Audio%20Processing",
|
| 177 |
"health": "/health",
|
| 178 |
"models-status": "/models-status"
|
| 179 |
}
|
|
|
|
| 190 |
status = {
|
| 191 |
"stt_model": "unknown",
|
| 192 |
"tts_engine": "gtts (free)",
|
| 193 |
+
"chatbot_model": "unknown",
|
| 194 |
+
"stance_model": "unknown",
|
| 195 |
+
"kpa_model": "unknown"
|
| 196 |
}
|
| 197 |
|
| 198 |
try:
|
| 199 |
+
from services.stt_service import STTService
|
| 200 |
+
stt_service = STTService()
|
| 201 |
+
status["stt_model"] = "loaded" if hasattr(stt_service, 'initialized') and stt_service.initialized else "not loaded"
|
| 202 |
except:
|
| 203 |
status["stt_model"] = "error"
|
| 204 |
|
| 205 |
try:
|
| 206 |
+
from services.chatbot_service import ChatbotService
|
| 207 |
+
chatbot_service = ChatbotService()
|
| 208 |
+
status["chatbot_model"] = "loaded" if hasattr(chatbot_service, 'initialized') and chatbot_service.initialized else "not loaded"
|
| 209 |
except:
|
| 210 |
status["chatbot_model"] = "error"
|
| 211 |
|
|
|
|
| 234 |
port=PORT,
|
| 235 |
reload=RELOAD,
|
| 236 |
log_level="info"
|
| 237 |
+
)
|
models/audio.py
CHANGED
|
@@ -1,44 +1,30 @@
|
|
| 1 |
from pydantic import BaseModel, Field
|
| 2 |
-
from typing import Optional
|
|
|
|
|
|
|
| 3 |
|
| 4 |
-
class
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
language: Optional[str] = Field(default="en", description="Language detected")
|
| 8 |
-
duration_seconds: Optional[float] = Field(None, description="Audio duration")
|
| 9 |
-
|
| 10 |
-
class Config:
|
| 11 |
-
json_schema_extra = {
|
| 12 |
-
"example": {
|
| 13 |
-
"text": "hello how are you",
|
| 14 |
-
"model_name": "whisper-base",
|
| 15 |
-
"language": "en",
|
| 16 |
-
"duration_seconds": 3.2
|
| 17 |
-
}
|
| 18 |
-
}
|
| 19 |
|
| 20 |
-
class
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
class ChatbotRequest(BaseModel):
|
| 27 |
-
text: str = Field(..., min_length=1, max_length=500, description="User input")
|
| 28 |
-
|
| 29 |
-
class Config:
|
| 30 |
-
json_schema_extra = {"example": {"text": "What is AI?"}}
|
| 31 |
|
| 32 |
class ChatbotResponse(BaseModel):
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
|
|
|
|
|
| 1 |
from pydantic import BaseModel, Field
|
| 2 |
+
from typing import Optional, List, Dict, Any
|
| 3 |
+
from enum import Enum
|
| 4 |
+
from datetime import datetime
|
| 5 |
|
| 6 |
+
class MessageType(str, Enum):
|
| 7 |
+
TEXT = "text"
|
| 8 |
+
AUDIO = "audio"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
|
| 10 |
+
class UserMessage(BaseModel):
|
| 11 |
+
message_id: str = Field(..., description="Unique message ID")
|
| 12 |
+
content: str = Field(..., description="Text content or audio base64")
|
| 13 |
+
message_type: MessageType = Field(..., description="Message type")
|
| 14 |
+
session_id: str = Field(..., description="User session ID")
|
| 15 |
+
timestamp: datetime = Field(default_factory=datetime.now)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
|
| 17 |
class ChatbotResponse(BaseModel):
|
| 18 |
+
response_id: str = Field(..., description="Unique response ID")
|
| 19 |
+
text_response: str = Field(..., description="Chatbot text response")
|
| 20 |
+
audio_response: Optional[str] = Field(None, description="Audio response in base64")
|
| 21 |
+
audio_url: Optional[str] = Field(None, description="Generated audio URL")
|
| 22 |
+
session_id: str = Field(..., description="User session ID")
|
| 23 |
+
timestamp: datetime = Field(default_factory=datetime.now)
|
| 24 |
+
|
| 25 |
+
class ChatSession(BaseModel):
|
| 26 |
+
session_id: str = Field(..., description="Session ID")
|
| 27 |
+
user_id: Optional[str] = Field(None, description="User ID")
|
| 28 |
+
created_at: datetime = Field(default_factory=datetime.now)
|
| 29 |
+
last_activity: datetime = Field(default_factory=datetime.now)
|
| 30 |
+
conversation_history: List[Dict[str, Any]] = Field(default_factory=list)
|
requirements.txt
CHANGED
|
@@ -9,14 +9,14 @@ pydantic==2.5.0
|
|
| 9 |
python-dotenv==1.0.0
|
| 10 |
torch>=2.0.0
|
| 11 |
transformers>=4.35.0
|
| 12 |
-
accelerate>=0.24.0
|
| 13 |
protobuf>=3.20.0
|
| 14 |
huggingface_hub>=0.19.0
|
| 15 |
python-multipart
|
| 16 |
google-genai>=0.4.0
|
| 17 |
-
gtts==2.5.1
|
| 18 |
requests==2.31.0
|
| 19 |
-
ffmpeg-python==0.2.0
|
| 20 |
-
librosa==0.10.1
|
| 21 |
soundfile==0.12.1
|
| 22 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
python-dotenv==1.0.0
|
| 10 |
torch>=2.0.0
|
| 11 |
transformers>=4.35.0
|
|
|
|
| 12 |
protobuf>=3.20.0
|
| 13 |
huggingface_hub>=0.19.0
|
| 14 |
python-multipart
|
| 15 |
google-genai>=0.4.0
|
|
|
|
| 16 |
requests==2.31.0
|
|
|
|
|
|
|
| 17 |
soundfile==0.12.1
|
| 18 |
+
gtts==2.3.2
|
| 19 |
+
SpeechRecognition==3.10.0
|
| 20 |
+
pyttsx3==2.90
|
| 21 |
+
accelerate>=0.20.0
|
| 22 |
+
coqui-tts==0.21.0
|
routes/audio.py
CHANGED
|
@@ -1,170 +1,86 @@
|
|
| 1 |
-
from fastapi import APIRouter, UploadFile, File,
|
| 2 |
-
from fastapi.responses import
|
| 3 |
-
import
|
| 4 |
-
import
|
| 5 |
-
from
|
| 6 |
-
from services.
|
| 7 |
-
from services.tts_service import generate_tts
|
| 8 |
-
from services.chatbot_service import get_chatbot_response, load_chatbot_model
|
| 9 |
-
from models.audio import STTResponse, TTSRequest, ChatbotRequest, ChatbotResponse
|
| 10 |
|
| 11 |
-
|
| 12 |
-
|
| 13 |
|
| 14 |
-
@router.
|
| 15 |
-
async def
|
| 16 |
-
|
| 17 |
-
|
|
|
|
|
|
|
|
|
|
| 18 |
try:
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
except Exception as e:
|
| 23 |
-
logger.error(f"⚠️ Model loading issues: {str(e)}")
|
| 24 |
-
|
| 25 |
-
@router.post("/tts")
|
| 26 |
-
async def tts(request: TTSRequest):
|
| 27 |
-
"""
|
| 28 |
-
Convert text to speech.
|
| 29 |
-
Returns MP3 audio file.
|
| 30 |
-
|
| 31 |
-
Example:
|
| 32 |
-
```
|
| 33 |
-
POST /audio/tts
|
| 34 |
-
{
|
| 35 |
-
"text": "Hello, welcome to voice chatbot"
|
| 36 |
-
}
|
| 37 |
-
```
|
| 38 |
-
"""
|
| 39 |
-
try:
|
| 40 |
-
logger.info(f"TTS Request: '{request.text}'")
|
| 41 |
-
audio_bytes = await generate_tts(request.text)
|
| 42 |
-
return StreamingResponse(
|
| 43 |
-
io.BytesIO(audio_bytes),
|
| 44 |
-
media_type="audio/mpeg",
|
| 45 |
-
headers={"Content-Disposition": "attachment; filename=output.mp3"}
|
| 46 |
-
)
|
| 47 |
-
except Exception as e:
|
| 48 |
-
logger.error(f"TTS Error: {str(e)}")
|
| 49 |
-
raise HTTPException(status_code=500, detail=str(e))
|
| 50 |
-
|
| 51 |
-
@router.post("/stt", response_model=STTResponse)
|
| 52 |
-
async def stt(file: UploadFile = File(...)):
|
| 53 |
-
"""
|
| 54 |
-
Convert audio file to text.
|
| 55 |
-
Supports: WAV, MP3, M4A
|
| 56 |
-
|
| 57 |
-
Example:
|
| 58 |
-
```
|
| 59 |
-
POST /audio/stt
|
| 60 |
-
File: audio.mp3
|
| 61 |
-
```
|
| 62 |
-
"""
|
| 63 |
-
if file.content_type not in ALLOWED_AUDIO_TYPES:
|
| 64 |
-
raise HTTPException(
|
| 65 |
-
status_code=400,
|
| 66 |
-
detail=f"Unsupported format. Allowed: {', '.join(ALLOWED_AUDIO_TYPES)}"
|
| 67 |
-
)
|
| 68 |
-
|
| 69 |
-
try:
|
| 70 |
-
logger.info(f"STT Request: {file.filename}")
|
| 71 |
-
audio_bytes = await file.read()
|
| 72 |
|
| 73 |
-
if
|
| 74 |
-
raise HTTPException(
|
| 75 |
-
status_code=400,
|
| 76 |
-
detail=f"File too large. Max: {MAX_AUDIO_SIZE / 1024 / 1024}MB"
|
| 77 |
-
)
|
| 78 |
|
| 79 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 80 |
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
|
| 84 |
-
|
| 85 |
-
|
| 86 |
-
|
| 87 |
-
raise
|
| 88 |
-
except Exception as e:
|
| 89 |
-
logger.error(f"STT Error: {str(e)}")
|
| 90 |
-
raise HTTPException(status_code=500, detail=str(e))
|
| 91 |
-
|
| 92 |
-
@router.post("/chatbot")
|
| 93 |
-
async def chatbot_voice(file: UploadFile = File(...)):
|
| 94 |
-
"""
|
| 95 |
-
Full voice chatbot flow: Audio → Text → Response → Audio
|
| 96 |
-
|
| 97 |
-
Example:
|
| 98 |
-
```
|
| 99 |
-
POST /audio/chatbot
|
| 100 |
-
File: user_voice.mp3
|
| 101 |
-
Returns: Response MP3 audio
|
| 102 |
-
```
|
| 103 |
-
"""
|
| 104 |
-
if file.content_type not in ALLOWED_AUDIO_TYPES:
|
| 105 |
-
raise HTTPException(
|
| 106 |
-
status_code=400,
|
| 107 |
-
detail=f"Unsupported format. Allowed: {', '.join(ALLOWED_AUDIO_TYPES)}"
|
| 108 |
)
|
| 109 |
-
|
| 110 |
-
try:
|
| 111 |
-
logger.info(f"Voice Chatbot: Processing {file.filename}")
|
| 112 |
-
|
| 113 |
-
audio_bytes = await file.read()
|
| 114 |
-
if len(audio_bytes) > MAX_AUDIO_SIZE:
|
| 115 |
-
raise HTTPException(
|
| 116 |
-
status_code=400,
|
| 117 |
-
detail=f"File too large. Max: {MAX_AUDIO_SIZE / 1024 / 1024}MB"
|
| 118 |
-
)
|
| 119 |
-
|
| 120 |
-
# Step 1: STT
|
| 121 |
-
logger.info("Step 1/3: Converting speech to text...")
|
| 122 |
-
user_text = await speech_to_text(audio_bytes, file.filename)
|
| 123 |
-
|
| 124 |
-
# Step 2: Chatbot response
|
| 125 |
-
logger.info("Step 2/3: Generating response...")
|
| 126 |
-
response_text = await get_chatbot_response(user_text)
|
| 127 |
-
|
| 128 |
-
# Step 3: TTS
|
| 129 |
-
logger.info("Step 3/3: Converting response to speech...")
|
| 130 |
-
audio_response = await generate_tts(response_text)
|
| 131 |
|
| 132 |
-
|
|
|
|
| 133 |
|
| 134 |
-
return
|
| 135 |
-
io.BytesIO(audio_response),
|
| 136 |
-
media_type="audio/mpeg",
|
| 137 |
-
headers={"Content-Disposition": "attachment; filename=response.mp3"}
|
| 138 |
-
)
|
| 139 |
|
| 140 |
-
except HTTPException:
|
| 141 |
-
raise
|
| 142 |
except Exception as e:
|
| 143 |
-
|
| 144 |
-
raise HTTPException(status_code=500, detail=str(e))
|
| 145 |
|
| 146 |
-
@router.post("/
|
| 147 |
-
async def
|
| 148 |
-
|
| 149 |
-
|
| 150 |
-
|
| 151 |
-
|
| 152 |
-
```
|
| 153 |
-
POST /audio/chatbot-text
|
| 154 |
-
{
|
| 155 |
-
"text": "What is artificial intelligence?"
|
| 156 |
-
}
|
| 157 |
-
```
|
| 158 |
-
"""
|
| 159 |
try:
|
| 160 |
-
|
| 161 |
-
|
| 162 |
|
| 163 |
-
|
| 164 |
-
|
| 165 |
-
|
| 166 |
-
|
|
|
|
| 167 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
| 168 |
except Exception as e:
|
| 169 |
-
|
| 170 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from fastapi import APIRouter, HTTPException, UploadFile, File, Form
|
| 2 |
+
from fastapi.responses import JSONResponse
|
| 3 |
+
import uuid
|
| 4 |
+
import base64
|
| 5 |
+
from models.audio import UserMessage, ChatbotResponse, MessageType
|
| 6 |
+
from services.chatbot_service import ChatbotService
|
|
|
|
|
|
|
|
|
|
| 7 |
|
| 8 |
+
router = APIRouter()
|
| 9 |
+
chatbot_service = ChatbotService()
|
| 10 |
|
| 11 |
+
@router.post("/chat/message", response_model=ChatbotResponse)
|
| 12 |
+
async def send_chat_message(
|
| 13 |
+
session_id: str = Form(...),
|
| 14 |
+
message_type: str = Form(...),
|
| 15 |
+
message: str = Form(None),
|
| 16 |
+
audio_file: UploadFile = File(None)
|
| 17 |
+
):
|
| 18 |
try:
|
| 19 |
+
# Validate input
|
| 20 |
+
if not message and not audio_file:
|
| 21 |
+
raise HTTPException(status_code=400, detail="Either message or audio file must be provided")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
|
| 23 |
+
if message_type == "audio" and not audio_file:
|
| 24 |
+
raise HTTPException(status_code=400, detail="Audio file required for audio messages")
|
|
|
|
|
|
|
|
|
|
| 25 |
|
| 26 |
+
# Process audio file if provided
|
| 27 |
+
content = ""
|
| 28 |
+
if audio_file:
|
| 29 |
+
audio_data = await audio_file.read()
|
| 30 |
+
content = base64.b64encode(audio_data).decode('utf-8')
|
| 31 |
+
else:
|
| 32 |
+
content = message
|
| 33 |
|
| 34 |
+
# Create user message
|
| 35 |
+
user_message = UserMessage(
|
| 36 |
+
message_id=str(uuid.uuid4()),
|
| 37 |
+
content=content,
|
| 38 |
+
message_type=MessageType(message_type),
|
| 39 |
+
session_id=session_id
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41 |
|
| 42 |
+
# Process through chatbot service
|
| 43 |
+
response = await chatbot_service.process_user_message(user_message)
|
| 44 |
|
| 45 |
+
return response
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
|
|
|
|
|
|
|
| 47 |
except Exception as e:
|
| 48 |
+
raise HTTPException(status_code=500, detail=f"Error processing message: {str(e)}")
|
|
|
|
| 49 |
|
| 50 |
+
@router.post("/chat/audio")
|
| 51 |
+
async def send_audio_message(
|
| 52 |
+
session_id: str = Form(...),
|
| 53 |
+
audio_file: UploadFile = File(...)
|
| 54 |
+
):
|
| 55 |
+
"""Endpoint specifically for audio messages"""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 56 |
try:
|
| 57 |
+
audio_data = await audio_file.read()
|
| 58 |
+
audio_base64 = base64.b64encode(audio_data).decode('utf-8')
|
| 59 |
|
| 60 |
+
user_message = UserMessage(
|
| 61 |
+
message_id=str(uuid.uuid4()),
|
| 62 |
+
content=audio_base64,
|
| 63 |
+
message_type=MessageType.AUDIO,
|
| 64 |
+
session_id=session_id
|
| 65 |
)
|
| 66 |
+
|
| 67 |
+
response = await chatbot_service.process_user_message(user_message)
|
| 68 |
+
return response
|
| 69 |
+
|
| 70 |
except Exception as e:
|
| 71 |
+
raise HTTPException(status_code=500, detail=f"Error processing audio: {str(e)}")
|
| 72 |
+
|
| 73 |
+
@router.get("/session/{session_id}/history")
|
| 74 |
+
async def get_session_history(session_id: str):
|
| 75 |
+
"""Get conversation history for a session"""
|
| 76 |
+
history = chatbot_service.get_session_history(session_id)
|
| 77 |
+
if not history:
|
| 78 |
+
raise HTTPException(status_code=404, detail="Session not found")
|
| 79 |
+
return history
|
| 80 |
+
|
| 81 |
+
@router.post("/session/new")
|
| 82 |
+
async def create_new_session():
|
| 83 |
+
"""Create a new chat session"""
|
| 84 |
+
session_id = str(uuid.uuid4())
|
| 85 |
+
chatbot_service._get_or_create_session(session_id)
|
| 86 |
+
return {"session_id": session_id, "message": "New session created"}
|
services/chatbot_service.py
CHANGED
|
@@ -1,69 +1,99 @@
|
|
| 1 |
-
import
|
| 2 |
-
|
| 3 |
-
import
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4 |
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
try:
|
| 12 |
-
logger.info("Loading DialoGPT chatbot model...")
|
| 13 |
-
chatbot_pipeline = pipeline(
|
| 14 |
-
"conversational",
|
| 15 |
-
model="microsoft/DialoGPT-medium",
|
| 16 |
-
device="cpu" # Use "cuda" if GPU available
|
| 17 |
-
)
|
| 18 |
-
logger.info("✓ Chatbot model loaded successfully")
|
| 19 |
-
except Exception as e:
|
| 20 |
-
logger.error(f"✗ Failed to load chatbot model: {str(e)}")
|
| 21 |
-
chatbot_pipeline = None
|
| 22 |
-
|
| 23 |
-
async def get_chatbot_response(user_text: str, user_id: str = "default") -> str:
|
| 24 |
-
"""
|
| 25 |
-
Generate chatbot response using DialoGPT.
|
| 26 |
-
Maintains conversation history per user.
|
| 27 |
-
"""
|
| 28 |
-
global chatbot_pipeline, conversation_history
|
| 29 |
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 35 |
|
| 36 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
|
| 38 |
-
#
|
| 39 |
-
|
| 40 |
-
|
|
|
|
|
|
|
|
|
|
| 41 |
|
| 42 |
-
#
|
| 43 |
-
|
| 44 |
-
conversation.add_user_input(user_text)
|
| 45 |
|
| 46 |
-
#
|
| 47 |
-
|
| 48 |
-
|
|
|
|
|
|
|
|
|
|
| 49 |
|
| 50 |
-
|
| 51 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 52 |
|
| 53 |
-
|
| 54 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 55 |
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import base64
|
| 2 |
+
import uuid
|
| 3 |
+
from typing import Optional, Dict, Any
|
| 4 |
+
from datetime import datetime
|
| 5 |
+
from models.audio import ChatbotResponse, UserMessage
|
| 6 |
+
from services.tts_service import SimpleTTSService # Use simple version
|
| 7 |
+
from services.stt_service import STTService # Use basic version
|
| 8 |
|
| 9 |
+
class ChatbotService:
|
| 10 |
+
def __init__(self):
|
| 11 |
+
self.tts_service = SimpleTTSService() # Use simple TTS
|
| 12 |
+
self.stt_service = STTService() # Use basic STT
|
| 13 |
+
self.sessions: Dict[str, Dict[str, Any]] = {}
|
| 14 |
+
self.initialized = False
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
|
| 16 |
+
async def initialize(self):
|
| 17 |
+
"""Initialize the chatbot service"""
|
| 18 |
+
await self.stt_service.initialize()
|
| 19 |
+
self.initialized = True
|
| 20 |
+
print("✓ Chatbot Service initialized")
|
| 21 |
+
|
| 22 |
+
async def process_user_message(self, user_message: UserMessage) -> ChatbotResponse:
|
| 23 |
+
# Update session
|
| 24 |
+
session = self._get_or_create_session(user_message.session_id)
|
| 25 |
|
| 26 |
+
# Process message based on type
|
| 27 |
+
if user_message.message_type == "audio":
|
| 28 |
+
# STT: Convert audio to text
|
| 29 |
+
text_input = await self.stt_service.transcribe_audio_base64(
|
| 30 |
+
user_message.content
|
| 31 |
+
)
|
| 32 |
+
else:
|
| 33 |
+
text_input = user_message.content
|
| 34 |
|
| 35 |
+
# Add to conversation history
|
| 36 |
+
session["conversation_history"].append({
|
| 37 |
+
"role": "user",
|
| 38 |
+
"content": text_input,
|
| 39 |
+
"timestamp": user_message.timestamp
|
| 40 |
+
})
|
| 41 |
|
| 42 |
+
# Generate chatbot response
|
| 43 |
+
chatbot_text = await self._generate_chatbot_response(text_input, session)
|
|
|
|
| 44 |
|
| 45 |
+
# TTS: Convert response to audio
|
| 46 |
+
try:
|
| 47 |
+
audio_base64 = await self.tts_service.text_to_speech_base64(chatbot_text)
|
| 48 |
+
except Exception as e:
|
| 49 |
+
print(f"TTS error: {e}")
|
| 50 |
+
audio_base64 = None
|
| 51 |
|
| 52 |
+
# Create response
|
| 53 |
+
response = ChatbotResponse(
|
| 54 |
+
response_id=str(uuid.uuid4()),
|
| 55 |
+
text_response=chatbot_text,
|
| 56 |
+
audio_response=audio_base64,
|
| 57 |
+
session_id=user_message.session_id
|
| 58 |
+
)
|
| 59 |
|
| 60 |
+
# Add response to history
|
| 61 |
+
session["conversation_history"].append({
|
| 62 |
+
"role": "assistant",
|
| 63 |
+
"content": chatbot_text,
|
| 64 |
+
"audio_response": audio_base64,
|
| 65 |
+
"timestamp": response.timestamp
|
| 66 |
+
})
|
| 67 |
|
| 68 |
+
return response
|
| 69 |
+
|
| 70 |
+
async def _generate_chatbot_response(self, user_input: str, session: Dict[str, Any]) -> str:
|
| 71 |
+
"""Chatbot response generation logic"""
|
| 72 |
+
# Simple response logic - replace with your actual chatbot model
|
| 73 |
+
user_input_lower = user_input.lower()
|
| 74 |
+
|
| 75 |
+
if "hello" in user_input_lower or "hi" in user_input_lower:
|
| 76 |
+
return "Hello! How can I assist you today?"
|
| 77 |
+
elif "time" in user_input_lower:
|
| 78 |
+
return f"The current time is {datetime.now().strftime('%H:%M')}"
|
| 79 |
+
elif "help" in user_input_lower:
|
| 80 |
+
return "I'm here to help you. You can ask me questions or request assistance."
|
| 81 |
+
elif "audio" in user_input_lower or "voice" in user_input_lower:
|
| 82 |
+
return "I can process both text and voice messages. Try sending me a voice message!"
|
| 83 |
+
else:
|
| 84 |
+
return f"I received your message: '{user_input}'. How can I assist you further?"
|
| 85 |
+
|
| 86 |
+
def _get_or_create_session(self, session_id: str) -> Dict[str, Any]:
|
| 87 |
+
if session_id not in self.sessions:
|
| 88 |
+
self.sessions[session_id] = {
|
| 89 |
+
"conversation_history": [],
|
| 90 |
+
"created_at": datetime.now(),
|
| 91 |
+
"last_activity": datetime.now()
|
| 92 |
+
}
|
| 93 |
+
else:
|
| 94 |
+
self.sessions[session_id]["last_activity"] = datetime.now()
|
| 95 |
+
|
| 96 |
+
return self.sessions[session_id]
|
| 97 |
+
|
| 98 |
+
def get_session_history(self, session_id: str) -> Optional[Dict[str, Any]]:
|
| 99 |
+
return self.sessions.get(session_id)
|
services/stt_service.py
CHANGED
|
@@ -1,66 +1,94 @@
|
|
| 1 |
-
import
|
|
|
|
| 2 |
import tempfile
|
| 3 |
import os
|
| 4 |
-
|
| 5 |
-
import
|
| 6 |
-
import numpy as np
|
| 7 |
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
def load_stt_model():
|
| 12 |
-
global stt_pipeline
|
| 13 |
-
try:
|
| 14 |
-
logger.info("Loading Whisper-base STT model...")
|
| 15 |
-
stt_pipeline = pipeline(
|
| 16 |
-
"automatic-speech-recognition",
|
| 17 |
-
model="openai/whisper-base",
|
| 18 |
-
device="cpu", # Use "cuda" if GPU available
|
| 19 |
-
chunk_length_s=30,
|
| 20 |
-
)
|
| 21 |
-
logger.info("✓ Whisper STT model loaded successfully")
|
| 22 |
-
except Exception as e:
|
| 23 |
-
logger.error(f"✗ Failed to load STT model: {str(e)}")
|
| 24 |
-
stt_pipeline = None
|
| 25 |
-
|
| 26 |
-
async def speech_to_text(audio_bytes: bytes, filename: str) -> str:
|
| 27 |
-
"""
|
| 28 |
-
Convert audio bytes to text using Whisper.
|
| 29 |
-
Handles WAV, MP3, M4A formats automatically.
|
| 30 |
-
"""
|
| 31 |
-
global stt_pipeline
|
| 32 |
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
|
| 41 |
-
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 45 |
|
| 46 |
try:
|
| 47 |
-
|
| 48 |
-
audio, sr = librosa.load(tmp_path, sr=16000)
|
| 49 |
|
| 50 |
-
|
| 51 |
-
|
| 52 |
-
|
| 53 |
|
| 54 |
-
|
| 55 |
-
|
| 56 |
|
| 57 |
-
|
| 58 |
-
return
|
| 59 |
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
except Exception as e:
|
| 65 |
-
logger.error(f"✗ STT Error: {str(e)}")
|
| 66 |
-
raise Exception(f"STT failed: {str(e)}")
|
|
|
|
| 1 |
+
import base64
|
| 2 |
+
import io
|
| 3 |
import tempfile
|
| 4 |
import os
|
| 5 |
+
import wave
|
| 6 |
+
import audioop
|
|
|
|
| 7 |
|
| 8 |
+
class STTService:
|
| 9 |
+
def __init__(self):
|
| 10 |
+
self.initialized = False
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
|
| 12 |
+
async def initialize(self):
|
| 13 |
+
"""Initialize STT service"""
|
| 14 |
+
# For now, we'll use a simple approach without external dependencies
|
| 15 |
+
self.initialized = True
|
| 16 |
+
print("✓ STT Service initialized (basic mode)")
|
| 17 |
+
|
| 18 |
+
async def transcribe_audio_base64(self, audio_base64: str, language: str = "en-US") -> str:
|
| 19 |
+
"""Transcribe base64 audio to text - SIMPLIFIED VERSION"""
|
| 20 |
+
try:
|
| 21 |
+
# Decode audio
|
| 22 |
+
audio_data = base64.b64decode(audio_base64)
|
| 23 |
+
|
| 24 |
+
# For now, return a placeholder since we don't have STT models configured
|
| 25 |
+
# In a real implementation, you would use Whisper, Vosk, or other STT models here
|
| 26 |
+
|
| 27 |
+
audio_info = await self._get_audio_info(audio_data)
|
| 28 |
+
return f"[Audio received: {audio_info}. STT service needs model configuration.]"
|
| 29 |
+
|
| 30 |
+
except Exception as e:
|
| 31 |
+
print(f"Transcription error: {e}")
|
| 32 |
+
return "Sorry, I couldn't process the audio message."
|
| 33 |
+
|
| 34 |
+
async def _get_audio_info(self, audio_data: bytes) -> str:
|
| 35 |
+
"""Get basic information about the audio file"""
|
| 36 |
+
try:
|
| 37 |
+
with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as temp_file:
|
| 38 |
+
temp_path = temp_file.name
|
| 39 |
+
temp_file.write(audio_data)
|
| 40 |
+
|
| 41 |
+
try:
|
| 42 |
+
with wave.open(temp_path, 'rb') as wav_file:
|
| 43 |
+
frames = wav_file.getnframes()
|
| 44 |
+
rate = wav_file.getframerate()
|
| 45 |
+
duration = frames / float(rate)
|
| 46 |
+
return f"Duration: {duration:.2f}s, Sample Rate: {rate}Hz"
|
| 47 |
+
except:
|
| 48 |
+
return f"Size: {len(audio_data)} bytes"
|
| 49 |
|
| 50 |
+
finally:
|
| 51 |
+
if os.path.exists(temp_path):
|
| 52 |
+
os.unlink(temp_path)
|
| 53 |
+
|
| 54 |
+
# Alternative STT service using Whisper if available
|
| 55 |
+
class WhisperSTTService:
|
| 56 |
+
def __init__(self):
|
| 57 |
+
self.model = None
|
| 58 |
+
self.initialized = False
|
| 59 |
+
|
| 60 |
+
async def initialize(self):
|
| 61 |
+
"""Initialize Whisper STT service"""
|
| 62 |
+
try:
|
| 63 |
+
import whisper
|
| 64 |
+
self.model = whisper.load_model("medium")
|
| 65 |
+
self.initialized = True
|
| 66 |
+
print("✓ Whisper STT Service initialized")
|
| 67 |
+
except ImportError:
|
| 68 |
+
print("⚠️ Whisper not available. Install with: pip install openai-whisper")
|
| 69 |
+
self.initialized = False
|
| 70 |
+
except Exception as e:
|
| 71 |
+
print(f"⚠️ Whisper initialization failed: {e}")
|
| 72 |
+
self.initialized = False
|
| 73 |
+
|
| 74 |
+
async def transcribe_audio_base64(self, audio_base64: str, language: str = "en") -> str:
|
| 75 |
+
"""Transcribe using Whisper"""
|
| 76 |
+
if not self.initialized:
|
| 77 |
+
return "STT service not available. Please install Whisper."
|
| 78 |
|
| 79 |
try:
|
| 80 |
+
audio_data = base64.b64decode(audio_base64)
|
|
|
|
| 81 |
|
| 82 |
+
with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as temp_file:
|
| 83 |
+
temp_path = temp_file.name
|
| 84 |
+
temp_file.write(audio_data)
|
| 85 |
|
| 86 |
+
result = self.model.transcribe(temp_path, language=language)
|
| 87 |
+
transcription = result["text"]
|
| 88 |
|
| 89 |
+
os.unlink(temp_path)
|
| 90 |
+
return transcription
|
| 91 |
|
| 92 |
+
except Exception as e:
|
| 93 |
+
print(f"Whisper transcription error: {e}")
|
| 94 |
+
return "Sorry, I couldn't transcribe the audio."
|
|
|
|
|
|
|
|
|
|
|
|
services/tts_service.py
CHANGED
|
@@ -1,49 +1,121 @@
|
|
| 1 |
-
import
|
| 2 |
import io
|
|
|
|
|
|
|
| 3 |
from gtts import gTTS
|
| 4 |
-
import
|
| 5 |
-
import wave
|
| 6 |
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
raise ValueError("Text must be between 1-500 characters")
|
| 17 |
|
| 18 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 19 |
|
| 20 |
-
#
|
| 21 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
|
| 23 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 24 |
audio_buffer = io.BytesIO()
|
| 25 |
tts.write_to_fp(audio_buffer)
|
| 26 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
|
| 28 |
-
|
| 29 |
-
|
| 30 |
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
"""Generate 1-second silent WAV file as fallback"""
|
| 38 |
-
sample_rate = 22050
|
| 39 |
-
duration = 1.0
|
| 40 |
-
silence = np.zeros(int(sample_rate * duration), dtype=np.int16)
|
| 41 |
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
|
| 46 |
-
|
| 47 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
|
| 49 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import base64
|
| 2 |
import io
|
| 3 |
+
import tempfile
|
| 4 |
+
import os
|
| 5 |
from gtts import gTTS
|
| 6 |
+
import pyttsx3
|
|
|
|
| 7 |
|
| 8 |
+
class TTSService:
|
| 9 |
+
def __init__(self):
|
| 10 |
+
self.models = {}
|
| 11 |
+
self._initialize_models()
|
| 12 |
+
|
| 13 |
+
def _initialize_models(self):
|
| 14 |
+
"""Initialize TTS models"""
|
| 15 |
+
# gTTS is our primary method (always available)
|
| 16 |
+
self.models["gtts"] = True
|
|
|
|
| 17 |
|
| 18 |
+
# Try to initialize pyttsx3 as fallback
|
| 19 |
+
try:
|
| 20 |
+
self.models["pyttsx3"] = pyttsx3.init()
|
| 21 |
+
print("✓ pyttsx3 TTS initialized")
|
| 22 |
+
except:
|
| 23 |
+
print("⚠️ pyttsx3 not available")
|
| 24 |
+
self.models["pyttsx3"] = None
|
| 25 |
|
| 26 |
+
# Coqui TTS is optional
|
| 27 |
+
self.models["coqui"] = self._initialize_coqui_tts()
|
| 28 |
+
|
| 29 |
+
def _initialize_coqui_tts(self):
|
| 30 |
+
"""Initialize Coqui TTS if available"""
|
| 31 |
+
try:
|
| 32 |
+
from TTS.api import TTS
|
| 33 |
+
tts_model = TTS(model_name="tts_models/en/ljspeech/tacotron2-DDC", progress_bar=False)
|
| 34 |
+
print("✓ Coqui TTS initialized")
|
| 35 |
+
return tts_model
|
| 36 |
+
except ImportError:
|
| 37 |
+
print("⚠️ Coqui TTS not available. Install with: pip install TTS")
|
| 38 |
+
return None
|
| 39 |
+
except Exception as e:
|
| 40 |
+
print(f"⚠️ Coqui TTS initialization failed: {e}")
|
| 41 |
+
return None
|
| 42 |
+
|
| 43 |
+
async def text_to_speech_base64(self, text: str, language: str = "en") -> str:
|
| 44 |
+
"""Convert text to base64 audio"""
|
| 45 |
+
# Try gTTS first (most reliable and free)
|
| 46 |
+
try:
|
| 47 |
+
return await self._gtts_to_base64(text, language)
|
| 48 |
+
except Exception as e:
|
| 49 |
+
print(f"gTTS error: {e}")
|
| 50 |
|
| 51 |
+
# Fallback to pyttsx3
|
| 52 |
+
try:
|
| 53 |
+
if self.models.get("pyttsx3"):
|
| 54 |
+
return await self._pyttsx3_to_base64(text)
|
| 55 |
+
except Exception as e:
|
| 56 |
+
print(f"pyttsx3 error: {e}")
|
| 57 |
+
|
| 58 |
+
# Final fallback to Coqui TTS
|
| 59 |
+
try:
|
| 60 |
+
if self.models.get("coqui"):
|
| 61 |
+
return await self._coqui_to_base64(text)
|
| 62 |
+
except Exception as e:
|
| 63 |
+
print(f"Coqui TTS error: {e}")
|
| 64 |
+
|
| 65 |
+
raise Exception("All TTS services failed")
|
| 66 |
+
|
| 67 |
+
async def _gtts_to_base64(self, text: str, language: str) -> str:
|
| 68 |
+
"""Convert using gTTS"""
|
| 69 |
+
tts = gTTS(text=text, lang=language, slow=False)
|
| 70 |
audio_buffer = io.BytesIO()
|
| 71 |
tts.write_to_fp(audio_buffer)
|
| 72 |
+
audio_buffer.seek(0)
|
| 73 |
+
return base64.b64encode(audio_buffer.getvalue()).decode('utf-8')
|
| 74 |
+
|
| 75 |
+
async def _pyttsx3_to_base64(self, text: str) -> str:
|
| 76 |
+
"""Convert using pyttsx3"""
|
| 77 |
+
engine = self.models["pyttsx3"]
|
| 78 |
+
with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as temp_file:
|
| 79 |
+
temp_path = temp_file.name
|
| 80 |
|
| 81 |
+
engine.save_to_file(text, temp_path)
|
| 82 |
+
engine.runAndWait()
|
| 83 |
|
| 84 |
+
with open(temp_path, 'rb') as audio_file:
|
| 85 |
+
audio_base64 = base64.b64encode(audio_file.read()).decode('utf-8')
|
| 86 |
+
|
| 87 |
+
# Cleanup
|
| 88 |
+
os.unlink(temp_path)
|
| 89 |
+
return audio_base64
|
|
|
|
|
|
|
|
|
|
|
|
|
| 90 |
|
| 91 |
+
async def _coqui_to_base64(self, text: str) -> str:
|
| 92 |
+
"""Convert using Coqui TTS"""
|
| 93 |
+
with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as temp_file:
|
| 94 |
+
temp_path = temp_file.name
|
| 95 |
+
|
| 96 |
+
self.models["coqui"].tts_to_file(text=text, file_path=temp_path)
|
| 97 |
+
|
| 98 |
+
with open(temp_path, 'rb') as audio_file:
|
| 99 |
+
audio_base64 = base64.b64encode(audio_file.read()).decode('utf-8')
|
| 100 |
+
|
| 101 |
+
# Cleanup
|
| 102 |
+
os.unlink(temp_path)
|
| 103 |
+
return audio_base64
|
| 104 |
+
|
| 105 |
+
# Simple TTS service that only uses gTTS (minimal dependencies)
|
| 106 |
+
class SimpleTTSService:
|
| 107 |
+
def __init__(self):
|
| 108 |
+
pass
|
| 109 |
|
| 110 |
+
async def text_to_speech_base64(self, text: str, language: str = "en") -> str:
|
| 111 |
+
"""Convert text to base64 audio using only gTTS"""
|
| 112 |
+
try:
|
| 113 |
+
tts = gTTS(text=text, lang=language, slow=False)
|
| 114 |
+
audio_buffer = io.BytesIO()
|
| 115 |
+
tts.write_to_fp(audio_buffer)
|
| 116 |
+
audio_buffer.seek(0)
|
| 117 |
+
return base64.b64encode(audio_buffer.getvalue()).decode('utf-8')
|
| 118 |
+
except Exception as e:
|
| 119 |
+
print(f"gTTS error: {e}")
|
| 120 |
+
# Return a placeholder audio or error message
|
| 121 |
+
return "TTS_ERROR_PLACEHOLDER"
|