Spaces:
Sleeping
Sleeping
File size: 14,912 Bytes
91d209c | 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 | """
Whisper-based Video Trimming and Frame Extraction
Uses OpenAI's Whisper to detect last spoken word and find optimal transition frames
"""
import re
from typing import List, Tuple, Optional
import tempfile
import os
try:
import whisper
from moviepy.editor import VideoFileClip
WHISPER_AVAILABLE = True
USE_SYSTEM_WHISPER = False
except ImportError:
# Try to use system Python's Whisper
import subprocess
import sys
SYSTEM_PYTHON = "/opt/anaconda3/bin/python"
if os.path.exists(SYSTEM_PYTHON):
try:
# Test if system Python has whisper
result = subprocess.run(
[SYSTEM_PYTHON, "-c", "import whisper; print('OK')"],
capture_output=True,
timeout=5
)
if result.returncode == 0:
WHISPER_AVAILABLE = True
USE_SYSTEM_WHISPER = True
else:
WHISPER_AVAILABLE = False
USE_SYSTEM_WHISPER = False
except:
WHISPER_AVAILABLE = False
USE_SYSTEM_WHISPER = False
else:
WHISPER_AVAILABLE = False
USE_SYSTEM_WHISPER = False
if not WHISPER_AVAILABLE:
print("β οΈ Whisper not available. Install with: pip install openai-whisper moviepy")
def normalize_text(text: str) -> str:
"""Normalize text by removing punctuation and converting to lowercase"""
return re.sub(r"[^\w\s]", "", text.lower().strip())
def transcribe_video(
video_path: str,
model_size: str = "base"
) -> Tuple[str, Optional[float]]:
"""
Transcribe video audio and find the last word timestamp.
Args:
video_path: Path to video file
model_size: Whisper model size (tiny, base, small, medium, large)
Returns:
Tuple of (full_transcription, last_word_end_time)
"""
if not WHISPER_AVAILABLE:
raise ImportError("Whisper not installed. Run: pip install openai-whisper moviepy")
# Use system Python if needed
if USE_SYSTEM_WHISPER:
return _transcribe_video_system(video_path, model_size)
print(f"π€ Loading Whisper model ({model_size})...")
model = whisper.load_model(model_size)
print("π€ Transcribing audio...")
result = model.transcribe(video_path, word_timestamps=True)
# Get full transcription text
full_text = result.get("text", "").strip()
# Get last word timestamp
segments = result.get("segments", [])
last_time = None
if segments:
# Find the end time of the last word
for seg in reversed(segments):
words = seg.get("words", [])
if words:
last_time = words[-1].get("end")
break
print(f"π Transcribed: \"{full_text[:100]}...\"" if len(full_text) > 100 else f"π Transcribed: \"{full_text}\"")
if last_time:
print(f"β
Last word ends at {last_time:.2f} seconds")
return full_text, last_time
def _transcribe_video_system(
video_path: str,
model_size: str = "base"
) -> Tuple[str, Optional[float]]:
"""Transcribe video using system Python's Whisper"""
import subprocess
import json
SYSTEM_PYTHON = "/opt/anaconda3/bin/python"
print(f"π€ Using system Whisper (model: {model_size})...")
# Create a temporary Python script file to avoid shell escaping issues
script_file = tempfile.NamedTemporaryFile(mode='w', suffix='.py', delete=False)
try:
script_file.write(f"""
import whisper
import json
import sys
video_path = {json.dumps(video_path)}
model_size = {json.dumps(model_size)}
model = whisper.load_model(model_size)
result = model.transcribe(video_path, word_timestamps=True)
# Extract transcription and last word timestamp
full_text = result.get("text", "").strip()
segments = result.get("segments", [])
last_time = None
if segments:
for seg in reversed(segments):
words = seg.get("words", [])
if words:
last_time = words[-1].get("end")
break
output = {{
"text": full_text,
"last_time": last_time
}}
print(json.dumps(output))
""")
script_file.close()
result = subprocess.run(
[SYSTEM_PYTHON, script_file.name],
capture_output=True,
text=True,
timeout=300 # 5 minute timeout
)
if result.returncode != 0:
raise Exception(f"Whisper transcription failed: {result.stderr}")
output = json.loads(result.stdout.strip())
full_text = output.get("text", "").strip()
last_time = output.get("last_time")
print(f"π Transcribed: \"{full_text[:100]}...\"" if len(full_text) > 100 else f"π Transcribed: \"{full_text}\"")
if last_time:
print(f"β
Last word ends at {last_time:.2f} seconds")
return full_text, last_time
except subprocess.TimeoutExpired:
raise Exception("Whisper transcription timed out")
except json.JSONDecodeError as e:
raise Exception(f"Failed to parse Whisper output: {str(e)}")
except Exception as e:
raise Exception(f"System Whisper error: {str(e)}")
finally:
# Clean up script file
try:
if os.path.exists(script_file.name):
os.remove(script_file.name)
except:
pass
def find_last_word_timestamp(
video_path: str,
script: str,
model_size: str = "base"
) -> Optional[float]:
"""
Find the timestamp of the last spoken word in the script
Args:
video_path: Path to video file
script: Expected script/dialogue
model_size: Whisper model size (tiny, base, small, medium, large)
Returns:
Timestamp (seconds) of last word, or None if not found
"""
if not WHISPER_AVAILABLE:
raise ImportError("Whisper not installed. Run: pip install openai-whisper moviepy")
# Use system Python if needed
if USE_SYSTEM_WHISPER:
return _find_last_word_timestamp_system(video_path, script, model_size)
print(f"π€ Loading Whisper model ({model_size})...")
model = whisper.load_model(model_size)
print("π€ Transcribing audio...")
result = model.transcribe(video_path, word_timestamps=True)
segments = result.get("segments", [])
if not segments:
raise ValueError("No speech segments detected in video")
# Find last word in script
script_clean = normalize_text(script)
script_words = script_clean.split()
if not script_words:
raise ValueError("Script is empty")
last_word = script_words[-1]
print(f"π Looking for last word: '{last_word}'")
# Search for last occurrence of that word in transcription
last_time = None
for seg in segments:
for word_info in seg.get("words", []):
word_text = normalize_text(word_info["word"])
if word_text == last_word:
last_time = word_info["end"]
if last_time is None:
# Fallback: try to find any word from the end of script
print(f"β οΈ Last word '{last_word}' not found, trying other words...")
for i in range(min(5, len(script_words))): # Try last 5 words
word_to_find = script_words[-(i+1)]
for seg in segments:
for word_info in seg.get("words", []):
word_text = normalize_text(word_info["word"])
if word_text == word_to_find:
last_time = word_info["end"]
print(f"β
Found '{word_to_find}' at {last_time:.2f}s instead")
break
if last_time:
break
if last_time:
break
if last_time:
print(f"β
Last spoken word ends at {last_time:.2f} seconds")
return last_time
def extract_post_speech_frames(
video_path: str,
script: str,
buffer_time: float = 0.3,
num_frames: int = 3,
model_size: str = "base"
) -> List[Tuple[float, str]]:
"""
Extract frames from the post-speech zone (after last spoken word)
Args:
video_path: Path to video file
script: Expected script/dialogue
buffer_time: Time after last word to start extracting (seconds)
num_frames: Number of frames to extract
model_size: Whisper model size
Returns:
List of (timestamp, base64_data_url) tuples
"""
if not WHISPER_AVAILABLE:
raise ImportError("Whisper not installed. Run: pip install openai-whisper moviepy")
# Find last word timestamp
last_word_time = find_last_word_timestamp(video_path, script, model_size)
if last_word_time is None:
raise ValueError("Could not find last spoken word in video")
# Get video duration
clip = VideoFileClip(video_path)
duration = clip.duration
clip.close()
# Calculate post-speech zone
post_speech_start = min(last_word_time + buffer_time, duration - 0.5)
post_speech_end = duration
print(f"π Post-speech zone: {post_speech_start:.2f}s to {post_speech_end:.2f}s")
# Calculate frame timestamps
available_time = post_speech_end - post_speech_start
if available_time < 0.1:
# Very little time, just use the end
timestamps = [duration - 0.1]
else:
# Distribute frames evenly in post-speech zone
if num_frames == 1:
timestamps = [post_speech_end - 0.1]
else:
step = available_time / (num_frames - 1)
timestamps = [post_speech_start + (i * step) for i in range(num_frames)]
# Extract frames
from utils.video_processor import extract_frame
frames = []
for i, timestamp in enumerate(timestamps):
print(f"πΈ Extracting frame at {timestamp:.2f}s...")
frame_data = extract_frame(video_path, timestamp, return_base64=True)
# Create label based on position
if i == 0 and len(timestamps) > 1:
label = "Right After Speech"
elif i == len(timestamps) - 1:
label = "Final Frame"
else:
label = f"Frame {i+1}"
frames.append((timestamp, frame_data, label))
return frames
def trim_video_to_last_word(
video_path: str,
script: str,
output_path: str,
padding: float = 0.5,
model_size: str = "base"
) -> str:
"""
Trim video to end shortly after the last spoken word
Args:
video_path: Input video path
script: Expected script/dialogue
output_path: Output video path
padding: Time to keep after last word (seconds)
model_size: Whisper model size
Returns:
Path to trimmed video
"""
if not WHISPER_AVAILABLE:
raise ImportError("Whisper not installed. Run: pip install openai-whisper moviepy")
# Find last word timestamp
last_word_time = find_last_word_timestamp(video_path, script, model_size)
if last_word_time is None:
raise ValueError("Could not find last spoken word in video")
# Calculate trim point
trim_time = last_word_time + padding
print(f"βοΈ Trimming video to {trim_time:.2f} seconds...")
# Use FFmpeg for trimming (more reliable than moviepy, especially with system Python)
import subprocess
# FFmpeg command to trim video
cmd = [
"ffmpeg",
"-i", video_path,
"-t", str(trim_time), # Duration to keep
"-c", "copy", # Copy codecs (fast, no re-encoding)
"-avoid_negative_ts", "make_zero",
"-y", # Overwrite output file
output_path
]
result = subprocess.run(cmd, capture_output=True, text=True)
if result.returncode != 0:
# If copy codec fails, try re-encoding
print("β οΈ Copy codec failed, re-encoding...")
cmd = [
"ffmpeg",
"-i", video_path,
"-t", str(trim_time),
"-c:v", "libx264",
"-c:a", "aac",
"-y",
output_path
]
result = subprocess.run(cmd, capture_output=True, text=True)
if result.returncode != 0:
raise Exception(f"FFmpeg trimming failed: {result.stderr}")
print(f"β
Trimmed video saved to: {output_path}")
return output_path
def _find_last_word_timestamp_system(
video_path: str,
script: str,
model_size: str = "base"
) -> Optional[float]:
"""Find last word timestamp using system Python"""
import subprocess
import json
import tempfile
SYSTEM_PYTHON = "/opt/anaconda3/bin/python"
print(f"π€ Using system Whisper (model: {model_size})...")
# Create temp file for JSON output
temp_json = tempfile.NamedTemporaryFile(mode='w', suffix='.json', delete=False)
temp_json.close()
try:
# Run whisper via system Python
cmd = [
SYSTEM_PYTHON, "-m", "whisper",
video_path,
"--model", model_size,
"--output_format", "json",
"--output_dir", os.path.dirname(temp_json.name),
"--word_timestamps", "True"
]
result = subprocess.run(cmd, capture_output=True, text=True, timeout=300)
if result.returncode != 0:
raise Exception(f"Whisper transcription failed: {result.stderr}")
# Find JSON file
base_name = os.path.splitext(os.path.basename(video_path))[0]
json_path = os.path.join(os.path.dirname(temp_json.name), f"{base_name}.json")
if not os.path.exists(json_path):
raise Exception(f"JSON output not found: {json_path}")
with open(json_path, 'r') as f:
transcription_data = json.load(f)
# Find last word
script_clean = normalize_text(script)
script_words = script_clean.split()
if not script_words:
return None
last_word = script_words[-1]
segments = transcription_data.get("segments", [])
last_time = None
for seg in segments:
for word_info in seg.get("words", []):
word_text = normalize_text(word_info.get("word", ""))
if word_text == last_word:
last_time = word_info.get("end", 0)
# Cleanup
try:
os.remove(json_path)
except:
pass
return last_time
finally:
try:
os.remove(temp_json.name)
except:
pass
def is_whisper_available() -> bool:
"""Check if Whisper is installed and available"""
return WHISPER_AVAILABLE
|