Spaces:
Sleeping
Sleeping
File size: 19,719 Bytes
6deff13 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 |
import os
import tempfile
import logging
import numpy as np
import torch
import torch.nn as nn
import whisper
import librosa
import asyncio
from typing import List, Dict, Any, Optional
from fastapi import FastAPI, UploadFile, File, HTTPException, Request, Form
from fastapi.middleware.cors import CORSMiddleware
from fastapi.responses import JSONResponse
from PIL import Image
from torchvision import transforms
from transformers import DistilBertTokenizerFast, DistilBertForSequenceClassification
from speechbrain.inference.classifiers import EncoderClassifier
import torchaudio
import json
# Setup logging
logging.basicConfig(
level=logging.INFO,
format="%(asctime)s [%(levelname)s] %(message)s",
)
logger = logging.getLogger(__name__)
class ModelManager:
"""Centralized model management for all ML models."""
_instance = None
def __new__(cls):
if cls._instance is None:
cls._instance = super(ModelManager, cls).__new__(cls)
cls._instance._initialized = False
return cls._instance
def __init__(self):
if self._initialized:
return
self._initialized = True
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
logger.info(f"Using device: {self.device}")
self.emotion_model = None
self.whisper_model = None
self.text_tokenizer = None
self.text_model = None
self.speechbrain_model = None
# Model paths
self.MODEL_PATHS = {
'whisper_model': 'base',
'text_model': 'emotion-distilbert-model',
'speechbrain_model': 'speechbrain/emotion-recognition-wav2vec2-IEMOCAP'
}
# Constants
self.EMOTIONS = ["Angry", "Disgust", "Fear", "Happy", "Sad", "Surprise", "Neutral"]
self.SAMPLE_RATE = 16000
self.TEXT_EMOTIONS = ["sadness", "joy", "love", "anger", "fear", "surprise"]
# SpeechBrain emotion mapping
self.SPEECHBRAIN_EMOTION_MAP = {
'neu': 'Neutral',
'hap': 'Happy',
'sad': 'Sad',
'ang': 'Angry',
'fea': 'Fear',
'dis': 'Disgust',
'sur': 'Surprise'
}
def load_all_models(self):
"""Load all required models."""
try:
logger.info("Starting to load all models...")
self._load_emotion_model()
self._load_whisper_model()
self._load_text_models()
self._load_speechbrain_model()
logger.info("All models loaded successfully!")
return True
except Exception as e:
logger.error(f"Error loading models: {str(e)}")
raise
def _load_emotion_model(self):
"""Use DeepFace for emotion recognition."""
try:
logger.info("Loading DeepFace for emotion recognition...")
from deepface import DeepFace
self.emotion_model = DeepFace
logger.info("DeepFace loaded successfully")
except Exception as e:
logger.error(f"Failed to initialize DeepFace: {str(e)}")
raise
def _load_whisper_model(self):
"""Load the Whisper speech-to-text model."""
try:
logger.info("Loading Whisper model...")
self.whisper_model = whisper.load_model(self.MODEL_PATHS['whisper_model'])
logger.info("Whisper model loaded successfully")
except Exception as e:
logger.error(f"Failed to load Whisper model: {str(e)}")
raise
def _load_text_models(self):
"""Load the text emotion classification model and tokenizer."""
try:
logger.info("Loading text emotion model...")
model_path = self.MODEL_PATHS['text_model']
# Try to load from local path first, then from HuggingFace Hub
if os.path.exists(model_path):
self.text_tokenizer = DistilBertTokenizerFast.from_pretrained(model_path)
self.text_model = DistilBertForSequenceClassification.from_pretrained(model_path)
else:
# Use a public emotion model from HuggingFace
logger.info("Local model not found, using HuggingFace model...")
self.text_tokenizer = DistilBertTokenizerFast.from_pretrained("bhadresh-savani/distilbert-base-uncased-emotion")
self.text_model = DistilBertForSequenceClassification.from_pretrained("bhadresh-savani/distilbert-base-uncased-emotion")
self.text_model.eval()
logger.info("Text models loaded successfully")
except Exception as e:
logger.error(f"Failed to load text models: {str(e)}")
raise
def _load_speechbrain_model(self):
"""Load SpeechBrain emotion recognition model."""
try:
logger.info("Loading SpeechBrain emotion recognition model...")
self.speechbrain_model = EncoderClassifier.from_hparams(
source=self.MODEL_PATHS['speechbrain_model'],
savedir="pretrained_models/emotion-recognition-wav2vec2-IEMOCAP",
run_opts={"device": "cpu"}
)
logger.info("SpeechBrain emotion recognition model loaded successfully")
except Exception as e:
logger.error(f"Failed to load SpeechBrain model: {str(e)}")
raise
def get_emotion_model(self):
if self.emotion_model is None:
self._load_emotion_model()
return self.emotion_model
def get_whisper_model(self):
if self.whisper_model is None:
self._load_whisper_model()
return self.whisper_model
def get_text_models(self):
if self.text_model is None or self.text_tokenizer is None:
self._load_text_models()
return self.text_tokenizer, self.text_model
def get_speechbrain_model(self):
if self.speechbrain_model is None:
self._load_speechbrain_model()
return self.speechbrain_model
# Initialize FastAPI app
app = FastAPI(title="Manan ML API - Emotion Recognition")
# CORS middleware
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
expose_headers=["*"]
)
# Initialize model manager
model_manager = ModelManager()
# Image transformation pipeline
transform = transforms.Compose([
transforms.Resize((224, 224)),
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
])
@app.on_event("startup")
async def startup_event():
"""Initialize all models when the application starts."""
try:
logger.info("Starting model initialization...")
model_manager.load_all_models()
logger.info("All models initialized successfully!")
except Exception as e:
logger.error(f"Failed to initialize models: {str(e)}")
# Don't raise - let the app start and load models on demand
@app.get("/")
async def root():
"""Health check endpoint."""
return {
"status": "running",
"message": "Manan ML API is running!",
"endpoints": [
"/pred_face - Face emotion prediction",
"/predict_audio_batch - Voice emotion prediction",
"/predict_text/ - Text emotion prediction"
]
}
@app.get("/health")
async def health_check():
"""Health check endpoint."""
return {"status": "healthy", "device": str(model_manager.device)}
# Helper function for SpeechBrain prediction
def predict_emotion_speechbrain(audio_path: str) -> Dict[str, Any]:
"""Predict emotion from audio using SpeechBrain."""
try:
speechbrain_model = model_manager.get_speechbrain_model()
signal, sr = torchaudio.load(audio_path)
if sr != 16000:
resampler = torchaudio.transforms.Resample(sr, 16000)
signal = resampler(signal)
if signal.dim() == 1:
signal = signal.unsqueeze(0)
elif signal.dim() == 3:
signal = signal.squeeze(1)
device = next(speechbrain_model.mods.wav2vec2.parameters()).device
signal = signal.to(device)
with torch.no_grad():
feats = speechbrain_model.mods.wav2vec2(signal)
pooled = speechbrain_model.mods.avg_pool(feats)
out = speechbrain_model.mods.output_mlp(pooled)
out_prob = speechbrain_model.hparams.softmax(out)
score, index = torch.max(out_prob, dim=-1)
predicted_emotion = speechbrain_model.hparams.label_encoder.decode_ndim(index.cpu())
if isinstance(predicted_emotion, list):
if isinstance(predicted_emotion[0], list):
emotion_key = str(predicted_emotion[0][0]).lower()[:3]
else:
emotion_key = str(predicted_emotion[0]).lower()[:3]
else:
emotion_key = str(predicted_emotion).lower()[:3]
emotion = model_manager.SPEECHBRAIN_EMOTION_MAP.get(emotion_key, 'Neutral')
probs = out_prob[0].detach().cpu().numpy()
if probs.ndim > 1:
probs = probs.flatten()
all_emotions = speechbrain_model.hparams.label_encoder.decode_ndim(
torch.arange(len(probs))
)
prob_dict = {}
for i in range(len(probs)):
if i < len(all_emotions):
if isinstance(all_emotions[i], list):
key = str(all_emotions[i][0]).lower()[:3]
else:
key = str(all_emotions[i]).lower()[:3]
emotion_name = model_manager.SPEECHBRAIN_EMOTION_MAP.get(key, f'emotion_{i}')
prob_dict[emotion_name] = float(probs[i])
confidence = float(score[0])
return {
'emotion': emotion,
'confidence': confidence,
'probabilities': prob_dict
}
except Exception as e:
logger.error(f"Error predicting emotion with SpeechBrain: {str(e)}")
raise
def transcribe_audio(audio_path: str) -> str:
"""Transcribe audio to text using Whisper."""
try:
result = model_manager.whisper_model.transcribe(audio_path)
return result["text"].strip()
except Exception as e:
logger.error(f"Error in audio transcription: {str(e)}")
return ""
# ============== API ENDPOINTS ==============
@app.post("/pred_face")
async def predict_face_emotion(
files: List[UploadFile] = File(...),
questions: str = Form(None)
):
"""Predict emotions from face images using DeepFace."""
from deepface import DeepFace
logger.info(f"Received {len(files)} files for face prediction")
if not files:
raise HTTPException(status_code=400, detail="No files provided")
temp_files = []
try:
questions_data = {}
question_count = 0
if questions:
try:
questions_data = json.loads(questions)
question_count = len(questions_data)
except json.JSONDecodeError:
raise HTTPException(status_code=400, detail="Invalid questions JSON format.")
else:
question_count = 3
questions_data = {str(i): {"text": f"Question {i+1}", "imageCount": 1} for i in range(question_count)}
question_files = {str(i): [] for i in range(question_count)}
for file in files:
if '_' in file.filename and file.filename.startswith('q'):
try:
q_idx = file.filename.split('_')[0][1:]
if q_idx in question_files:
question_files[q_idx].append(file)
except Exception as e:
logger.warning(f"Skipping file {file.filename}: {e}")
results = []
for q_idx, q_files in question_files.items():
if not q_files:
results.append({
"emotion": "Unknown",
"probabilities": {e: 0.0 for e in model_manager.EMOTIONS}
})
continue
probs_list = []
for file in q_files:
try:
with tempfile.NamedTemporaryFile(delete=False, suffix=".jpg") as tmp:
content = await file.read()
tmp.write(content)
temp_path = tmp.name
temp_files.append(temp_path)
analysis = DeepFace.analyze(
img_path=temp_path,
actions=['emotion'],
enforce_detection=False,
silent=True
)
if isinstance(analysis, list):
analysis = analysis[0]
emotion_scores = analysis.get('emotion', {})
dominant_emotion = analysis.get('dominant_emotion', 'neutral')
normalized_probs = {}
for emo in model_manager.EMOTIONS:
key = emo.lower()
normalized_probs[emo] = emotion_scores.get(key, 0.0) / 100.0
probs_list.append(normalized_probs)
except Exception as e:
logger.error(f"Error processing {file.filename}: {e}")
if probs_list:
avg_probs = {}
for emo in model_manager.EMOTIONS:
avg_probs[emo] = sum(p.get(emo, 0) for p in probs_list) / len(probs_list)
dominant_emotion = max(avg_probs, key=avg_probs.get)
results.append({
"emotion": dominant_emotion,
"probabilities": avg_probs
})
else:
results.append({
"emotion": "Unknown",
"probabilities": {e: 0.0 for e in model_manager.EMOTIONS}
})
return results
except Exception as e:
logger.error(f"Error in face emotion prediction: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
finally:
for file_path in temp_files:
try:
if os.path.exists(file_path):
os.remove(file_path)
except Exception as e:
logger.warning(f"Failed to delete temp file {file_path}: {e}")
@app.post("/predict_audio_batch")
async def predict_audio_batch(files: List[UploadFile] = File(...)):
"""Predict emotions from multiple audio files using SpeechBrain."""
logger.info(f"Received {len(files)} audio files for prediction")
if not files:
raise HTTPException(status_code=400, detail="No audio files provided")
temp_files = []
results = []
try:
for file in files:
try:
with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp:
content = await file.read()
tmp.write(content)
temp_path = tmp.name
temp_files.append(temp_path)
prediction = predict_emotion_speechbrain(temp_path)
results.append(prediction)
logger.info(f"Predicted emotion for {file.filename}: {prediction['emotion']}")
except Exception as e:
logger.error(f"Error processing {file.filename}: {e}")
results.append({
'emotion': 'Unknown',
'confidence': 0.0,
'probabilities': {},
'error': str(e)
})
return {'status': 'success', 'results': results}
except Exception as e:
logger.error(f"Error in audio batch prediction: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
finally:
for file_path in temp_files:
try:
if os.path.exists(file_path):
os.remove(file_path)
except Exception as e:
logger.warning(f"Failed to delete temp file {file_path}: {e}")
@app.post("/predict_text/")
async def predict_text_emotion(files: List[UploadFile] = File(...)):
"""Transcribe audio and predict text emotion."""
logger.info(f"Received {len(files)} audio files for text prediction")
if not files:
raise HTTPException(status_code=400, detail="No audio files provided")
temp_files = []
results = []
try:
tokenizer, text_model = model_manager.get_text_models()
whisper_model = model_manager.get_whisper_model()
for file in files:
try:
with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp:
content = await file.read()
tmp.write(content)
temp_path = tmp.name
temp_files.append(temp_path)
# Transcribe
transcription = whisper_model.transcribe(temp_path)
transcript = transcription["text"].strip()
logger.info(f"Transcribed: {transcript}")
if not transcript:
results.append({
'transcript': '',
'emotion': 'neutral',
'confidence': 0.0,
'probabilities': {}
})
continue
# Predict emotion from text
inputs = tokenizer(
transcript,
return_tensors="pt",
truncation=True,
max_length=128,
padding=True
)
with torch.no_grad():
outputs = text_model(**inputs)
probs = torch.softmax(outputs.logits, dim=1)[0]
# Get emotion labels
emotion_labels = model_manager.TEXT_EMOTIONS
if hasattr(text_model.config, 'id2label'):
emotion_labels = [text_model.config.id2label[i] for i in range(len(probs))]
prob_dict = {emotion_labels[i]: float(probs[i]) for i in range(len(probs))}
predicted_idx = torch.argmax(probs).item()
predicted_emotion = emotion_labels[predicted_idx]
confidence = float(probs[predicted_idx])
results.append({
'transcript': transcript,
'emotion': predicted_emotion,
'confidence': confidence,
'probabilities': prob_dict
})
except Exception as e:
logger.error(f"Error processing {file.filename}: {e}")
results.append({
'transcript': '',
'emotion': 'unknown',
'confidence': 0.0,
'error': str(e)
})
return results
except Exception as e:
logger.error(f"Error in text prediction: {str(e)}")
raise HTTPException(status_code=500, detail=str(e))
finally:
for file_path in temp_files:
try:
if os.path.exists(file_path):
os.remove(file_path)
except Exception as e:
logger.warning(f"Failed to delete temp file {file_path}: {e}")
if __name__ == "__main__":
import uvicorn
uvicorn.run(app, host="0.0.0.0", port=7860)
|