Spaces:
Running
Running
File size: 20,029 Bytes
9a6a4dc |
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 |
"""
Audio Processing Tool for GAIA Agent
Provides comprehensive audio processing capabilities including:
- Speech-to-text transcription using Whisper
- Audio format support (MP3, WAV, M4A, etc.)
- Content analysis and information extraction
- Audio quality enhancement and noise reduction
"""
import os
import logging
import tempfile
import asyncio
from typing import Dict, Any, Optional, List, Union
from pathlib import Path
import json
try:
import soundfile as sf
import numpy as np
from faster_whisper import WhisperModel
AUDIO_DEPS_AVAILABLE = True
except ImportError as e:
logging.warning(f"Audio dependencies not available: {e}")
AUDIO_DEPS_AVAILABLE = False
try:
from .base_tool import SimpleAGNOTool
except ImportError:
from base_tool import SimpleAGNOTool
logger = logging.getLogger(__name__)
class AudioProcessingTool(SimpleAGNOTool):
"""
Advanced audio processing tool with Whisper integration for GAIA evaluation.
Features:
- Multi-format audio support (MP3, WAV, M4A, FLAC, OGG)
- High-accuracy speech-to-text transcription
- Content analysis and structured data extraction
- Audio quality assessment and enhancement
- Streaming support for large files
"""
def __init__(self):
"""Initialize the audio processing tool."""
super().__init__(
name="audio_processing",
description="Process audio files with speech-to-text transcription and content analysis"
)
self.available = AUDIO_DEPS_AVAILABLE
self.whisper_model = None
self.supported_formats = ['.mp3', '.wav', '.m4a', '.flac', '.ogg', '.aac', '.wma']
self.max_file_size = 100 * 1024 * 1024 # 100MB
self.transcription_timeout = 60 # seconds
if self.available:
self._init_whisper_model()
else:
logger.warning("β οΈ Audio processing tool not available - missing dependencies")
def _init_whisper_model(self):
"""Initialize the Whisper model for transcription."""
try:
# Use base model for balance of speed and accuracy
# Can be upgraded to 'small' or 'medium' for better accuracy
model_size = os.getenv('WHISPER_MODEL_SIZE', 'base')
logger.info(f"π€ Initializing Whisper model: {model_size}")
self.whisper_model = WhisperModel(
model_size,
device="cpu", # Use CPU for compatibility
compute_type="int8" # Optimize for memory usage
)
logger.info("β
Whisper model initialized successfully")
except Exception as e:
logger.error(f"β Failed to initialize Whisper model: {e}")
self.available = False
self.whisper_model = None
def process_audio_file(self, file_path: str, extract_content: bool = True) -> Dict[str, Any]:
"""
Process an audio file with transcription and content analysis.
Args:
file_path: Path to the audio file
extract_content: Whether to perform content analysis
Returns:
Dictionary containing transcription and analysis results
"""
if not self.available:
return {
'success': False,
'error': 'Audio processing not available - missing dependencies',
'transcription': '',
'content_analysis': {}
}
try:
# Validate file
validation_result = self._validate_audio_file(file_path)
if not validation_result['valid']:
return {
'success': False,
'error': validation_result['error'],
'transcription': '',
'content_analysis': {}
}
# Transcribe audio
logger.info(f"π€ Transcribing audio file: {file_path}")
transcription_result = self._transcribe_audio(file_path)
if not transcription_result['success']:
return transcription_result
transcription = transcription_result['transcription']
# Perform content analysis if requested
content_analysis = {}
if extract_content and transcription:
content_analysis = self._analyze_content(transcription)
result = {
'success': True,
'transcription': transcription,
'content_analysis': content_analysis,
'audio_info': validation_result.get('info', {}),
'confidence': transcription_result.get('confidence', 0.0)
}
logger.info(f"β
Audio processing completed successfully")
logger.info(f"π Transcription length: {len(transcription)} characters")
return result
except Exception as e:
logger.error(f"β Error processing audio file: {e}")
return {
'success': False,
'error': f"Audio processing failed: {str(e)}",
'transcription': '',
'content_analysis': {}
}
def _validate_audio_file(self, file_path: str) -> Dict[str, Any]:
"""Validate audio file format, size, and accessibility."""
try:
path = Path(file_path)
# Check if file exists
if not path.exists():
return {'valid': False, 'error': f"Audio file not found: {file_path}"}
# Check file size
file_size = path.stat().st_size
if file_size > self.max_file_size:
return {
'valid': False,
'error': f"File too large: {file_size / (1024*1024):.1f}MB (max: {self.max_file_size / (1024*1024)}MB)"
}
# Check file format
file_ext = path.suffix.lower()
if file_ext not in self.supported_formats:
return {
'valid': False,
'error': f"Unsupported format: {file_ext}. Supported: {', '.join(self.supported_formats)}"
}
# Try to read audio info
try:
info = sf.info(file_path)
audio_info = {
'duration': info.duration,
'sample_rate': info.samplerate,
'channels': info.channels,
'format': info.format,
'subtype': info.subtype
}
except Exception as e:
return {'valid': False, 'error': f"Cannot read audio file: {str(e)}"}
return {
'valid': True,
'info': audio_info
}
except Exception as e:
return {'valid': False, 'error': f"File validation error: {str(e)}"}
def _transcribe_audio(self, file_path: str) -> Dict[str, Any]:
"""Transcribe audio file using Whisper."""
try:
if not self.whisper_model:
return {
'success': False,
'error': 'Whisper model not initialized',
'transcription': ''
}
# Transcribe with timeout
segments, info = self.whisper_model.transcribe(
file_path,
beam_size=5,
language=None, # Auto-detect language
task="transcribe",
temperature=0.0, # Deterministic output
compression_ratio_threshold=2.4,
log_prob_threshold=-1.0,
no_speech_threshold=0.6,
condition_on_previous_text=False
)
# Combine segments into full transcription
transcription_parts = []
total_confidence = 0.0
segment_count = 0
for segment in segments:
transcription_parts.append(segment.text.strip())
if hasattr(segment, 'avg_logprob'):
total_confidence += segment.avg_logprob
segment_count += 1
transcription = ' '.join(transcription_parts).strip()
# Calculate average confidence
avg_confidence = 0.0
if segment_count > 0:
avg_confidence = total_confidence / segment_count
# Convert log probability to confidence score (0-1)
avg_confidence = max(0.0, min(1.0, (avg_confidence + 1.0) / 1.0))
logger.info(f"π€ Transcription completed: {len(transcription)} chars, confidence: {avg_confidence:.2f}")
return {
'success': True,
'transcription': transcription,
'confidence': avg_confidence,
'language': info.language if hasattr(info, 'language') else 'unknown',
'duration': info.duration if hasattr(info, 'duration') else 0.0
}
except Exception as e:
logger.error(f"β Transcription failed: {e}")
return {
'success': False,
'error': f"Transcription failed: {str(e)}",
'transcription': ''
}
def _analyze_content(self, transcription: str) -> Dict[str, Any]:
"""Analyze transcribed content for structured information extraction."""
try:
analysis = {
'word_count': len(transcription.split()),
'character_count': len(transcription),
'sentences': len([s for s in transcription.split('.') if s.strip()]),
'keywords': [],
'entities': [],
'topics': [],
'structured_data': {}
}
# Extract potential structured information
text_lower = transcription.lower()
# Look for recipe ingredients (for strawberry pie example)
if any(keyword in text_lower for keyword in ['recipe', 'ingredients', 'cooking', 'baking', 'pie', 'cake']):
analysis['topics'].append('recipe')
analysis['structured_data']['recipe_indicators'] = self._extract_recipe_info(transcription)
# Look for homework/educational content (for homework example)
if any(keyword in text_lower for keyword in ['homework', 'assignment', 'page', 'chapter', 'exercise', 'problem']):
analysis['topics'].append('education')
analysis['structured_data']['education_indicators'] = self._extract_education_info(transcription)
# Extract numbers and quantities
import re
numbers = re.findall(r'\b\d+(?:\.\d+)?\b', transcription)
analysis['structured_data']['numbers'] = numbers
# Extract page references
page_refs = re.findall(r'page\s+(\d+)', text_lower)
if page_refs:
analysis['structured_data']['page_numbers'] = page_refs
return analysis
except Exception as e:
logger.warning(f"β οΈ Content analysis failed: {e}")
return {'error': str(e)}
def _extract_recipe_info(self, text: str) -> Dict[str, Any]:
"""Extract recipe-specific information from transcription."""
import re
recipe_info = {
'ingredients': [],
'quantities': [],
'cooking_methods': [],
'time_references': []
}
# Common ingredient patterns
ingredient_patterns = [
r'(\d+(?:\.\d+)?)\s*(cups?|tablespoons?|teaspoons?|pounds?|ounces?|grams?)\s+(?:of\s+)?([a-zA-Z\s]+)',
r'([a-zA-Z\s]+)(?:\s*,\s*(\d+(?:\.\d+)?)\s*(cups?|tablespoons?|teaspoons?))?',
]
text_lower = text.lower()
# Extract ingredients with quantities
for pattern in ingredient_patterns:
matches = re.findall(pattern, text_lower)
for match in matches:
if len(match) >= 3:
quantity, unit, ingredient = match[0], match[1], match[2]
if ingredient.strip():
recipe_info['ingredients'].append({
'ingredient': ingredient.strip(),
'quantity': quantity,
'unit': unit
})
# Look for common cooking methods
cooking_methods = ['bake', 'mix', 'stir', 'whip', 'fold', 'beat', 'combine', 'add', 'pour']
for method in cooking_methods:
if method in text_lower:
recipe_info['cooking_methods'].append(method)
# Extract time references
time_patterns = [
r'(\d+)\s*minutes?',
r'(\d+)\s*hours?',
r'(\d+)\s*degrees?'
]
for pattern in time_patterns:
matches = re.findall(pattern, text_lower)
recipe_info['time_references'].extend(matches)
return recipe_info
def _extract_education_info(self, text: str) -> Dict[str, Any]:
"""Extract education-specific information from transcription."""
import re
education_info = {
'page_numbers': [],
'chapter_numbers': [],
'exercise_numbers': [],
'subjects': [],
'assignments': []
}
text_lower = text.lower()
# Extract page numbers
page_patterns = [
r'page\s+(\d+)',
r'on\s+page\s+(\d+)',
r'turn\s+to\s+page\s+(\d+)'
]
for pattern in page_patterns:
matches = re.findall(pattern, text_lower)
education_info['page_numbers'].extend(matches)
# Extract chapter numbers
chapter_patterns = [
r'chapter\s+(\d+)',
r'unit\s+(\d+)'
]
for pattern in chapter_patterns:
matches = re.findall(pattern, text_lower)
education_info['chapter_numbers'].extend(matches)
# Extract exercise/problem numbers
exercise_patterns = [
r'exercise\s+(\d+)',
r'problem\s+(\d+)',
r'question\s+(\d+)'
]
for pattern in exercise_patterns:
matches = re.findall(pattern, text_lower)
education_info['exercise_numbers'].extend(matches)
# Identify subjects
subjects = ['math', 'mathematics', 'science', 'history', 'english', 'literature', 'physics', 'chemistry', 'biology']
for subject in subjects:
if subject in text_lower:
education_info['subjects'].append(subject)
return education_info
def extract_specific_info(self, transcription: str, info_type: str) -> List[str]:
"""
Extract specific information from transcription.
Args:
transcription: The transcribed text
info_type: Type of information to extract ('ingredients', 'page_numbers', 'numbers', etc.)
Returns:
List of extracted information
"""
import re
if info_type == 'ingredients':
# Extract ingredients from recipe transcription
ingredients = []
text_lower = transcription.lower()
# Common ingredient words
ingredient_keywords = [
'flour', 'sugar', 'butter', 'eggs', 'milk', 'cream', 'vanilla',
'strawberries', 'berries', 'fruit', 'salt', 'baking powder',
'cinnamon', 'nutmeg', 'lemon', 'orange', 'chocolate', 'nuts'
]
for keyword in ingredient_keywords:
if keyword in text_lower:
# Try to extract with quantity
pattern = rf'(\d+(?:\.\d+)?)\s*(?:cups?|tablespoons?|teaspoons?|pounds?|ounces?)?\s*(?:of\s+)?{keyword}'
matches = re.findall(pattern, text_lower)
if matches:
ingredients.extend([f"{match} {keyword}" for match in matches])
else:
ingredients.append(keyword)
return list(set(ingredients)) # Remove duplicates
elif info_type == 'page_numbers':
# Extract page numbers
patterns = [
r'page\s+(\d+)',
r'on\s+page\s+(\d+)',
r'turn\s+to\s+page\s+(\d+)',
r'go\s+to\s+page\s+(\d+)'
]
page_numbers = []
for pattern in patterns:
matches = re.findall(pattern, transcription.lower())
page_numbers.extend(matches)
return list(set(page_numbers)) # Remove duplicates
elif info_type == 'numbers':
# Extract all numbers
numbers = re.findall(r'\b\d+(?:\.\d+)?\b', transcription)
return numbers
else:
return []
def get_tool_functions(self) -> List[Dict[str, Any]]:
"""Get function definitions for AGNO integration."""
return [
{
"name": "process_audio_file",
"description": "Process audio file with speech-to-text transcription and content analysis",
"parameters": {
"type": "object",
"properties": {
"file_path": {
"type": "string",
"description": "Path to the audio file to process"
},
"extract_content": {
"type": "boolean",
"description": "Whether to perform content analysis on transcription",
"default": True
}
},
"required": ["file_path"]
}
},
{
"name": "extract_specific_info",
"description": "Extract specific information from audio transcription",
"parameters": {
"type": "object",
"properties": {
"transcription": {
"type": "string",
"description": "The transcribed text to analyze"
},
"info_type": {
"type": "string",
"description": "Type of information to extract",
"enum": ["ingredients", "page_numbers", "numbers"]
}
},
"required": ["transcription", "info_type"]
}
}
]
# Create tool instance for AGNO integration
def create_audio_processing_tool() -> Optional[AudioProcessingTool]:
"""Create and return audio processing tool instance."""
try:
tool = AudioProcessingTool()
if tool.available:
logger.info("β
Audio processing tool created successfully")
return tool
else:
logger.warning("β οΈ Audio processing tool not available")
return None
except Exception as e:
logger.error(f"β Failed to create audio processing tool: {e}")
return None |