Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
|
@@ -234,55 +234,23 @@ class VibeVoiceDemo:
|
|
| 234 |
|
| 235 |
|
| 236 |
def load_example_scripts(self):
|
| 237 |
-
"""Load example scripts from the text_examples directory."""
|
| 238 |
examples_dir = os.path.join(os.path.dirname(__file__), "text_examples")
|
| 239 |
self.example_scripts = []
|
| 240 |
-
|
| 241 |
-
# Check if text_examples directory exists
|
| 242 |
if not os.path.exists(examples_dir):
|
| 243 |
-
print(f"Warning: text_examples directory not found at {examples_dir}")
|
| 244 |
return
|
| 245 |
-
|
| 246 |
-
# Get all .txt files in the text_examples directory
|
| 247 |
-
txt_files = sorted([f for f in os.listdir(examples_dir)
|
| 248 |
-
if f.lower().endswith('.txt') and os.path.isfile(os.path.join(examples_dir, f))])
|
| 249 |
-
|
| 250 |
-
for txt_file in txt_files:
|
| 251 |
-
file_path = os.path.join(examples_dir, txt_file)
|
| 252 |
-
|
| 253 |
-
import re
|
| 254 |
-
# Check if filename contains a time pattern like "45min", "90min", etc.
|
| 255 |
-
time_pattern = re.search(r'(\d+)min', txt_file.lower())
|
| 256 |
-
if time_pattern:
|
| 257 |
-
minutes = int(time_pattern.group(1))
|
| 258 |
-
if minutes > 15:
|
| 259 |
-
print(f"Skipping {txt_file}: duration {minutes} minutes exceeds 15-minute limit")
|
| 260 |
-
continue
|
| 261 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 262 |
try:
|
| 263 |
-
with open(
|
| 264 |
script_content = f.read().strip()
|
| 265 |
-
|
| 266 |
-
|
| 267 |
-
|
| 268 |
-
|
| 269 |
-
if not script_content:
|
| 270 |
-
continue
|
| 271 |
-
|
| 272 |
-
# Parse the script to determine number of speakers
|
| 273 |
-
num_speakers = self._get_num_speakers_from_script(script_content)
|
| 274 |
-
|
| 275 |
-
# Add to examples list as [num_speakers, script_content]
|
| 276 |
-
self.example_scripts.append([num_speakers, script_content])
|
| 277 |
-
print(f"Loaded example: {txt_file} with {num_speakers} speakers")
|
| 278 |
-
|
| 279 |
except Exception as e:
|
| 280 |
-
print(f"Error loading
|
| 281 |
-
|
| 282 |
-
if self.example_scripts:
|
| 283 |
-
print(f"Successfully loaded {len(self.example_scripts)} example scripts")
|
| 284 |
-
else:
|
| 285 |
-
print("No example scripts were loaded")
|
| 286 |
|
| 287 |
|
| 288 |
def convert_to_16_bit_wav(data):
|
|
@@ -295,202 +263,7 @@ def convert_to_16_bit_wav(data):
|
|
| 295 |
|
| 296 |
|
| 297 |
def create_demo_interface(demo_instance: VibeVoiceDemo):
|
| 298 |
-
custom_css = """
|
| 299 |
-
.gradio-container {
|
| 300 |
-
background: linear-gradient(135deg, #f8fafc 0%, #e2e8f0 100%);
|
| 301 |
-
font-family: 'SF Pro Display', -apple-system, BlinkMacSystemFont, sans-serif;
|
| 302 |
-
}
|
| 303 |
-
|
| 304 |
-
/* Header styling */
|
| 305 |
-
.main-header {
|
| 306 |
-
background: linear-gradient(90deg, #667eea 0%, #764ba2 100%);
|
| 307 |
-
padding: 2rem;
|
| 308 |
-
border-radius: 20px;
|
| 309 |
-
margin-bottom: 2rem;
|
| 310 |
-
text-align: center;
|
| 311 |
-
box-shadow: 0 10px 40px rgba(102, 126, 234, 0.3);
|
| 312 |
-
}
|
| 313 |
-
|
| 314 |
-
.main-header h1 {
|
| 315 |
-
color: white;
|
| 316 |
-
font-size: 2.5rem;
|
| 317 |
-
font-weight: 700;
|
| 318 |
-
margin: 0;
|
| 319 |
-
text-shadow: 0 2px 4px rgba(0,0,0,0.3);
|
| 320 |
-
}
|
| 321 |
-
|
| 322 |
-
.main-header p {
|
| 323 |
-
color: rgba(255,255,255,0.9);
|
| 324 |
-
font-size: 1.1rem;
|
| 325 |
-
margin: 0.5rem 0 0 0;
|
| 326 |
-
}
|
| 327 |
-
|
| 328 |
-
/* Card styling */
|
| 329 |
-
.settings-card, .generation-card {
|
| 330 |
-
background: rgba(255, 255, 255, 0.8);
|
| 331 |
-
backdrop-filter: blur(10px);
|
| 332 |
-
border: 1px solid rgba(226, 232, 240, 0.8);
|
| 333 |
-
border-radius: 16px;
|
| 334 |
-
padding: 1.5rem;
|
| 335 |
-
margin-bottom: 1rem;
|
| 336 |
-
box-shadow: 0 8px 32px rgba(0, 0, 0, 0.1);
|
| 337 |
-
}
|
| 338 |
-
|
| 339 |
-
/* Speaker selection styling */
|
| 340 |
-
.speaker-grid {
|
| 341 |
-
display: grid;
|
| 342 |
-
gap: 1rem;
|
| 343 |
-
margin-bottom: 1rem;
|
| 344 |
-
}
|
| 345 |
-
|
| 346 |
-
.speaker-item {
|
| 347 |
-
background: linear-gradient(135deg, #e2e8f0 0%, #cbd5e1 100%);
|
| 348 |
-
border: 1px solid rgba(148, 163, 184, 0.4);
|
| 349 |
-
border-radius: 12px;
|
| 350 |
-
padding: 1rem;
|
| 351 |
-
color: #374151;
|
| 352 |
-
font-weight: 500;
|
| 353 |
-
}
|
| 354 |
-
|
| 355 |
-
/* Streaming indicator */
|
| 356 |
-
.streaming-indicator {
|
| 357 |
-
display: inline-block;
|
| 358 |
-
width: 10px;
|
| 359 |
-
height: 10px;
|
| 360 |
-
background: #22c55e;
|
| 361 |
-
border-radius: 50%;
|
| 362 |
-
margin-right: 8px;
|
| 363 |
-
animation: pulse 1.5s infinite;
|
| 364 |
-
}
|
| 365 |
-
|
| 366 |
-
@keyframes pulse {
|
| 367 |
-
0% { opacity: 1; transform: scale(1); }
|
| 368 |
-
50% { opacity: 0.5; transform: scale(1.1); }
|
| 369 |
-
100% { opacity: 1; transform: scale(1); }
|
| 370 |
-
}
|
| 371 |
-
|
| 372 |
-
/* Queue status styling */
|
| 373 |
-
.queue-status {
|
| 374 |
-
background: linear-gradient(135deg, #f0f9ff 0%, #e0f2fe 100%);
|
| 375 |
-
border: 1px solid rgba(14, 165, 233, 0.3);
|
| 376 |
-
border-radius: 8px;
|
| 377 |
-
padding: 0.75rem;
|
| 378 |
-
margin: 0.5rem 0;
|
| 379 |
-
text-align: center;
|
| 380 |
-
font-size: 0.9rem;
|
| 381 |
-
color: #0369a1;
|
| 382 |
-
}
|
| 383 |
-
|
| 384 |
-
.generate-btn {
|
| 385 |
-
background: linear-gradient(135deg, #059669 0%, #0d9488 100%);
|
| 386 |
-
border: none;
|
| 387 |
-
border-radius: 12px;
|
| 388 |
-
padding: 1rem 2rem;
|
| 389 |
-
color: white;
|
| 390 |
-
font-weight: 600;
|
| 391 |
-
font-size: 1.1rem;
|
| 392 |
-
box-shadow: 0 4px 20px rgba(5, 150, 105, 0.4);
|
| 393 |
-
transition: all 0.3s ease;
|
| 394 |
-
}
|
| 395 |
-
|
| 396 |
-
.generate-btn:hover {
|
| 397 |
-
transform: translateY(-2px);
|
| 398 |
-
box-shadow: 0 6px 25px rgba(5, 150, 105, 0.6);
|
| 399 |
-
}
|
| 400 |
-
|
| 401 |
-
.stop-btn {
|
| 402 |
-
background: linear-gradient(135deg, #ef4444 0%, #dc2626 100%);
|
| 403 |
-
border: none;
|
| 404 |
-
border-radius: 12px;
|
| 405 |
-
padding: 1rem 2rem;
|
| 406 |
-
color: white;
|
| 407 |
-
font-weight: 600;
|
| 408 |
-
font-size: 1.1rem;
|
| 409 |
-
box-shadow: 0 4px 20px rgba(239, 68, 68, 0.4);
|
| 410 |
-
transition: all 0.3s ease;
|
| 411 |
-
}
|
| 412 |
-
|
| 413 |
-
.stop-btn:hover {
|
| 414 |
-
transform: translateY(-2px);
|
| 415 |
-
box-shadow: 0 6px 25px rgba(239, 68, 68, 0.6);
|
| 416 |
-
}
|
| 417 |
-
|
| 418 |
-
/* Audio player styling */
|
| 419 |
-
.audio-output {
|
| 420 |
-
background: linear-gradient(135deg, #f1f5f9 0%, #e2e8f0 100%);
|
| 421 |
-
border-radius: 16px;
|
| 422 |
-
padding: 1.5rem;
|
| 423 |
-
border: 1px solid rgba(148, 163, 184, 0.3);
|
| 424 |
-
}
|
| 425 |
-
|
| 426 |
-
.complete-audio-section {
|
| 427 |
-
margin-top: 1rem;
|
| 428 |
-
padding: 1rem;
|
| 429 |
-
background: linear-gradient(135deg, #f0fdf4 0%, #dcfce7 100%);
|
| 430 |
-
border: 1px solid rgba(34, 197, 94, 0.3);
|
| 431 |
-
border-radius: 12px;
|
| 432 |
-
}
|
| 433 |
-
|
| 434 |
-
/* Text areas */
|
| 435 |
-
.script-input, .log-output {
|
| 436 |
-
background: rgba(255, 255, 255, 0.9) !important;
|
| 437 |
-
border: 1px solid rgba(148, 163, 184, 0.4) !important;
|
| 438 |
-
border-radius: 12px !important;
|
| 439 |
-
color: #1e293b !important;
|
| 440 |
-
font-family: 'JetBrains Mono', monospace !important;
|
| 441 |
-
}
|
| 442 |
-
|
| 443 |
-
.script-input::placeholder {
|
| 444 |
-
color: #64748b !important;
|
| 445 |
-
}
|
| 446 |
-
|
| 447 |
-
/* Sliders */
|
| 448 |
-
.slider-container {
|
| 449 |
-
background: rgba(248, 250, 252, 0.8);
|
| 450 |
-
border: 1px solid rgba(226, 232, 240, 0.6);
|
| 451 |
-
border-radius: 8px;
|
| 452 |
-
padding: 1rem;
|
| 453 |
-
margin: 0.5rem 0;
|
| 454 |
-
}
|
| 455 |
-
|
| 456 |
-
/* Labels and text */
|
| 457 |
-
.gradio-container label {
|
| 458 |
-
color: #374151 !important;
|
| 459 |
-
font-weight: 600 !important;
|
| 460 |
-
}
|
| 461 |
-
|
| 462 |
-
.gradio-container .markdown {
|
| 463 |
-
color: #1f2937 !important;
|
| 464 |
-
}
|
| 465 |
-
|
| 466 |
-
/* Responsive design */
|
| 467 |
-
@media (max-width: 768px) {
|
| 468 |
-
.main-header h1 { font-size: 2rem; }
|
| 469 |
-
.settings-card, .generation-card { padding: 1rem; }
|
| 470 |
-
}
|
| 471 |
-
|
| 472 |
-
/* Random example button styling - more subtle professional color */
|
| 473 |
-
.random-btn {
|
| 474 |
-
background: linear-gradient(135deg, #64748b 0%, #475569 100%);
|
| 475 |
-
border: none;
|
| 476 |
-
border-radius: 12px;
|
| 477 |
-
padding: 1rem 1.5rem;
|
| 478 |
-
color: white;
|
| 479 |
-
font-weight: 600;
|
| 480 |
-
font-size: 1rem;
|
| 481 |
-
box-shadow: 0 4px 20px rgba(100, 116, 139, 0.3);
|
| 482 |
-
transition: all 0.3s ease;
|
| 483 |
-
display: inline-flex;
|
| 484 |
-
align-items: center;
|
| 485 |
-
gap: 0.5rem;
|
| 486 |
-
}
|
| 487 |
-
|
| 488 |
-
.random-btn:hover {
|
| 489 |
-
transform: translateY(-2px);
|
| 490 |
-
box-shadow: 0 6px 25px rgba(100, 116, 139, 0.4);
|
| 491 |
-
background: linear-gradient(135deg, #475569 0%, #334155 100%);
|
| 492 |
-
}
|
| 493 |
-
"""
|
| 494 |
|
| 495 |
with gr.Blocks(
|
| 496 |
title="VibeVoice - AI Podcast Generator",
|
|
|
|
| 234 |
|
| 235 |
|
| 236 |
def load_example_scripts(self):
|
|
|
|
| 237 |
examples_dir = os.path.join(os.path.dirname(__file__), "text_examples")
|
| 238 |
self.example_scripts = []
|
|
|
|
|
|
|
| 239 |
if not os.path.exists(examples_dir):
|
|
|
|
| 240 |
return
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 241 |
|
| 242 |
+
txt_files = sorted(
|
| 243 |
+
[f for f in os.listdir(examples_dir) if f.lower().endswith('.txt')]
|
| 244 |
+
)
|
| 245 |
+
for txt_file in txt_files:
|
| 246 |
try:
|
| 247 |
+
with open(os.path.join(examples_dir, txt_file), 'r', encoding='utf-8') as f:
|
| 248 |
script_content = f.read().strip()
|
| 249 |
+
if script_content:
|
| 250 |
+
num_speakers = self._infer_num_speakers_from_script(script_content)
|
| 251 |
+
self.example_scripts.append([num_speakers, script_content])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 252 |
except Exception as e:
|
| 253 |
+
print(f"Error loading {txt_file}: {e}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 254 |
|
| 255 |
|
| 256 |
def convert_to_16_bit_wav(data):
|
|
|
|
| 263 |
|
| 264 |
|
| 265 |
def create_demo_interface(demo_instance: VibeVoiceDemo):
|
| 266 |
+
custom_css = """ """
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 267 |
|
| 268 |
with gr.Blocks(
|
| 269 |
title="VibeVoice - AI Podcast Generator",
|