Spaces:
Sleeping
Sleeping
File size: 8,845 Bytes
c5c9261 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 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 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 |
from fastapi import FastAPI, HTTPException, Header, Depends, Query
from fastapi.responses import JSONResponse, FileResponse
from fastapi.staticfiles import StaticFiles
from fastapi.middleware.cors import CORSMiddleware
from pydantic import BaseModel, Field
from typing import Optional, Literal, Dict, Any, List
from app.core.audio import decode_base64_audio, preprocess_audio
from app.core.model import voice_detector
from app.core.explanation import generate_explanation
from app.config import settings
import uvicorn
import logging
import os
# Setup Logger
logger = logging.getLogger("api")
app = FastAPI(
title="VoiceGuard — AI Voice Detection API",
description=(
"World-class deepfake audio detection API. "
"Combines neural model inference with multi-analyzer forensic analysis "
"to detect AI-generated speech with maximum accuracy."
),
version="3.0.0"
)
# CORS Security
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
# Mount Static Files (for the Web UI)
static_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "static")
if os.path.exists(static_path):
app.mount("/static", StaticFiles(directory=static_path), name="static")
# Custom Error Handlers (Hackathon Specification)
@app.exception_handler(HTTPException)
async def http_exception_handler(request, exc):
return JSONResponse(
status_code=exc.status_code,
content={"status": "error", "message": exc.detail}
)
@app.exception_handler(Exception)
async def general_exception_handler(request, exc):
logger.error(f"Unhandled exception: {str(exc)}")
return JSONResponse(
status_code=500,
content={"status": "error", "message": "Internal server error"}
)
# ===============================================================
# Request / Response Models
# ===============================================================
class AudioRequest(BaseModel):
language: Literal["Tamil", "English", "Hindi", "Malayalam", "Telugu"] = Field(
..., description="Language of the audio"
)
audioFormat: Literal["mp3", "wav"] = Field(
..., description="Audio format (mp3 or wav)"
)
audioBase64: str = Field(
..., description="Base64-encoded audio file"
)
class ForensicDetail(BaseModel):
score: float = Field(..., description="Analyzer score (0=human, 1=AI)")
verdict: str = Field(..., description="Analyzer verdict")
artifacts_found: List[str] = Field(default_factory=list)
details: Dict[str, Any] = Field(default_factory=dict)
class NeuralModelDetail(BaseModel):
score: float = Field(..., description="Neural model AI probability")
verdict: str = Field(..., description="Neural model verdict")
segments_analyzed: int = Field(..., description="Number of audio segments analyzed")
per_segment_scores: List[float] = Field(default_factory=list)
class AudioProfileResponse(BaseModel):
duration_sec: float
snr_db: float
clipping_detected: bool
silence_ratio: float
rms_energy: float
sample_rate: int
class DetectionResponse(BaseModel):
status: Literal["success", "error"] = Field(default="success")
language: str = Field(..., description="Language of the audio")
classification: Literal["AI_GENERATED", "HUMAN"] = Field(
..., description="Voice classification result"
)
confidenceScore: float = Field(
..., ge=0.0, le=1.0, description="Confidence score (0.0–1.0)"
)
explanation: str = Field(
..., description="Evidence-based explanation for the decision"
)
# Extended fields
inferenceTimeMs: Optional[float] = Field(
None, description="Total inference time in milliseconds"
)
analyzersAgree: Optional[bool] = Field(
None, description="Whether neural model and forensic analyzers agree"
)
forensics: Optional[Dict[str, Any]] = Field(
None, description="Detailed forensic analysis (when detailed=true)"
)
audioProfile: Optional[AudioProfileResponse] = Field(
None, description="Audio quality profile (when detailed=true)"
)
artifactsSummary: Optional[List[str]] = Field(
None, description="List of all detected artifacts (when detailed=true)"
)
class ErrorResponse(BaseModel):
status: Literal["error"] = "error"
message: str
# ===============================================================
# Authentication
# ===============================================================
def verify_api_key(
x_api_key_lower: str = Header(None, alias="x-api-key"),
x_api_key_upper: str = Header(None, alias="X-API-Key")
):
x_api_key = x_api_key_lower or x_api_key_upper
if not settings.API_KEY:
return True
if not x_api_key:
raise HTTPException(
status_code=401,
detail="API key is required. Please provide x-api-key header."
)
if x_api_key != settings.API_KEY:
raise HTTPException(status_code=403, detail="Invalid API key")
return True
# ===============================================================
# Routes
# ===============================================================
@app.get("/")
def read_root():
index_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "static", "index.html")
if os.path.exists(index_path):
return FileResponse(index_path)
return {
"status": "healthy",
"message": "VoiceGuard AI Voice Detection API v3.0",
"authentication": "enabled" if settings.API_KEY else "disabled"
}
@app.get("/test")
def test_console():
"""Serve the Base64 Testing Console."""
test_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "static", "test.html")
if os.path.exists(test_path):
return FileResponse(test_path)
return {"error": "test.html not found"}
@app.get("/health")
def health_check():
return {
"status": "healthy",
"version": "3.0.0",
"engine": "multi-analyzer forensic detection",
"model": settings.MODEL_NAME,
"analyzers": ["neural_model", "spectral_analysis", "temporal_analysis",
"formant_analysis", "artifact_detection"],
"authentication": "enabled" if settings.API_KEY else "disabled"
}
@app.post("/detect", response_model=DetectionResponse)
async def detect_voice(
request: AudioRequest,
authenticated: bool = Depends(verify_api_key),
detailed: bool = Query(False, description="Return full forensic breakdown")
):
"""
🎙️ World-Class AI Voice Detection
Analyzes audio using a multi-stage pipeline:
1. **Neural Model** — Wav2Vec2 deepfake classifier (multi-segment inference)
2. **Spectral Analyzer** — Detects unnatural spectral patterns
3. **Temporal Analyzer** — Catches robotic timing and uniform pauses
4. **Formant Analyzer** — Identifies synthetic formant transitions
5. **Artifact Detector** — Finds phase discontinuities and click artifacts
Results are fused for maximum accuracy with evidence-based explanations.
"""
try:
# 1. Decode Base64
audio_file = decode_base64_audio(request.audioBase64)
# 2. Preprocess Audio (now returns profile too)
audio_array, audio_profile = preprocess_audio(audio_file)
# 3. Full Detection Pipeline (neural + forensics + fusion)
result = voice_detector.predict(
audio_array,
audio_profile=audio_profile,
detailed=detailed,
)
# 4. Generate Evidence-Based Explanation
explanation = generate_explanation(result)
# 5. Build Response
response = DetectionResponse(
status="success",
language=request.language,
classification=result["classification"],
confidenceScore=round(result["confidence"], 2),
explanation=explanation,
inferenceTimeMs=result.get("inference_time_ms"),
analyzersAgree=result.get("analyzers_agree"),
)
# Include detailed forensics if requested
if detailed:
response.forensics = result.get("forensics")
response.artifactsSummary = result.get("artifacts_summary")
if audio_profile:
response.audioProfile = AudioProfileResponse(**audio_profile)
return response
except HTTPException as he:
raise he
except Exception as e:
logger.error(f"Error processing request: {str(e)}")
raise HTTPException(
status_code=500,
detail="Internal server error processing the audio"
)
if __name__ == "__main__":
uvicorn.run("app.main:app", host="0.0.0.0", port=8000, reload=True)
|