Spaces:
Running on Zero
Running on Zero
Remove stale style text input
Browse files
app.py
CHANGED
|
@@ -248,7 +248,6 @@ def _random_text() -> str:
|
|
| 248 |
@spaces.GPU
|
| 249 |
def synthesize(
|
| 250 |
text: str,
|
| 251 |
-
style_text: str,
|
| 252 |
language: str,
|
| 253 |
speaker_label: str,
|
| 254 |
emotion_label: str,
|
|
@@ -262,7 +261,6 @@ def synthesize(
|
|
| 262 |
|
| 263 |
Args:
|
| 264 |
text: The text to synthesize.
|
| 265 |
-
style_text: Optional style reference text (falls back to ``text``).
|
| 266 |
language: Language code (e.g. ``"en"``).
|
| 267 |
speaker_label: Speaker name from the dropdown.
|
| 268 |
emotion_label: Emotion name from the dropdown.
|
|
@@ -276,11 +274,6 @@ def synthesize(
|
|
| 276 |
A tuple of (wav_filepath, stats_json).
|
| 277 |
"""
|
| 278 |
clean_text = normalize_client_text(text, max_chars=MAX_TEXT_CHARS)
|
| 279 |
-
clean_style_text = (
|
| 280 |
-
normalize_client_text(style_text, max_chars=MAX_TEXT_CHARS)
|
| 281 |
-
if style_text and style_text.strip()
|
| 282 |
-
else ""
|
| 283 |
-
)
|
| 284 |
clean_language = normalize_language(language)
|
| 285 |
if speaker_label not in SPEAKER_IDS:
|
| 286 |
raise gr.Error("Choose a valid speaker.")
|
|
@@ -309,7 +302,7 @@ def synthesize(
|
|
| 309 |
audio_top_k=clean_top_k if clean_top_k > 0 else None,
|
| 310 |
audio_repetition_penalty=clean_rep_pen,
|
| 311 |
n_vq_for_inference=32,
|
| 312 |
-
style_text=
|
| 313 |
style_emotion_id=emotion_lookup[emotion_label],
|
| 314 |
style_energy=clean_energy,
|
| 315 |
)
|
|
@@ -373,12 +366,6 @@ with gr.Blocks(
|
|
| 373 |
lines=4,
|
| 374 |
max_length=MAX_TEXT_CHARS,
|
| 375 |
)
|
| 376 |
-
style_text = gr.Textbox(
|
| 377 |
-
label="Style text (optional)",
|
| 378 |
-
placeholder="Leave empty to use the same text as input.",
|
| 379 |
-
lines=2,
|
| 380 |
-
max_length=MAX_TEXT_CHARS,
|
| 381 |
-
)
|
| 382 |
with gr.Row():
|
| 383 |
randomize = gr.Button("Random Text")
|
| 384 |
generate = gr.Button("Generate", variant="primary")
|
|
@@ -430,17 +417,17 @@ with gr.Blocks(
|
|
| 430 |
|
| 431 |
gr.Examples(
|
| 432 |
examples=[
|
| 433 |
-
["I wasn't expecting that to work, but now I'm kind of excited.", "
|
| 434 |
DEFAULT_SPEAKER, "Neutral", 0.5, 0.8, 0.92, 0, 1.05],
|
| 435 |
-
["You have no idea how happy that made me.", "
|
| 436 |
DEFAULT_SPEAKER, "Happy", 0.8, 0.8, 0.92, 0, 1.05],
|
| 437 |
-
["I'm trying to stay calm, but that was way closer than I expected.", "
|
| 438 |
DEFAULT_SPEAKER, "Fearful", 0.3, 0.8, 0.92, 0, 1.05],
|
| 439 |
-
["Okay, deep breath. We can fix this if we take it one step at a time.", "
|
| 440 |
DEFAULT_SPEAKER, "Calm", 0.4, 0.8, 0.92, 0, 1.05],
|
| 441 |
],
|
| 442 |
inputs=[
|
| 443 |
-
text,
|
| 444 |
audio_temperature, audio_top_p, audio_top_k, audio_repetition_penalty,
|
| 445 |
],
|
| 446 |
outputs=[audio_output, stats],
|
|
@@ -453,7 +440,7 @@ with gr.Blocks(
|
|
| 453 |
generate.click(
|
| 454 |
synthesize,
|
| 455 |
inputs=[
|
| 456 |
-
text,
|
| 457 |
audio_temperature, audio_top_p, audio_top_k, audio_repetition_penalty,
|
| 458 |
],
|
| 459 |
outputs=[audio_output, stats],
|
|
@@ -462,4 +449,4 @@ with gr.Blocks(
|
|
| 462 |
|
| 463 |
if __name__ == "__main__":
|
| 464 |
demo.queue(default_concurrency_limit=1, max_size=20)
|
| 465 |
-
demo.launch(mcp_server=True, show_error=False)
|
|
|
|
| 248 |
@spaces.GPU
|
| 249 |
def synthesize(
|
| 250 |
text: str,
|
|
|
|
| 251 |
language: str,
|
| 252 |
speaker_label: str,
|
| 253 |
emotion_label: str,
|
|
|
|
| 261 |
|
| 262 |
Args:
|
| 263 |
text: The text to synthesize.
|
|
|
|
| 264 |
language: Language code (e.g. ``"en"``).
|
| 265 |
speaker_label: Speaker name from the dropdown.
|
| 266 |
emotion_label: Emotion name from the dropdown.
|
|
|
|
| 274 |
A tuple of (wav_filepath, stats_json).
|
| 275 |
"""
|
| 276 |
clean_text = normalize_client_text(text, max_chars=MAX_TEXT_CHARS)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 277 |
clean_language = normalize_language(language)
|
| 278 |
if speaker_label not in SPEAKER_IDS:
|
| 279 |
raise gr.Error("Choose a valid speaker.")
|
|
|
|
| 302 |
audio_top_k=clean_top_k if clean_top_k > 0 else None,
|
| 303 |
audio_repetition_penalty=clean_rep_pen,
|
| 304 |
n_vq_for_inference=32,
|
| 305 |
+
style_text=clean_text,
|
| 306 |
style_emotion_id=emotion_lookup[emotion_label],
|
| 307 |
style_energy=clean_energy,
|
| 308 |
)
|
|
|
|
| 366 |
lines=4,
|
| 367 |
max_length=MAX_TEXT_CHARS,
|
| 368 |
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 369 |
with gr.Row():
|
| 370 |
randomize = gr.Button("Random Text")
|
| 371 |
generate = gr.Button("Generate", variant="primary")
|
|
|
|
| 417 |
|
| 418 |
gr.Examples(
|
| 419 |
examples=[
|
| 420 |
+
["I wasn't expecting that to work, but now I'm kind of excited.", "en",
|
| 421 |
DEFAULT_SPEAKER, "Neutral", 0.5, 0.8, 0.92, 0, 1.05],
|
| 422 |
+
["You have no idea how happy that made me.", "en",
|
| 423 |
DEFAULT_SPEAKER, "Happy", 0.8, 0.8, 0.92, 0, 1.05],
|
| 424 |
+
["I'm trying to stay calm, but that was way closer than I expected.", "en",
|
| 425 |
DEFAULT_SPEAKER, "Fearful", 0.3, 0.8, 0.92, 0, 1.05],
|
| 426 |
+
["Okay, deep breath. We can fix this if we take it one step at a time.", "en",
|
| 427 |
DEFAULT_SPEAKER, "Calm", 0.4, 0.8, 0.92, 0, 1.05],
|
| 428 |
],
|
| 429 |
inputs=[
|
| 430 |
+
text, language, speaker, emotion, style_energy,
|
| 431 |
audio_temperature, audio_top_p, audio_top_k, audio_repetition_penalty,
|
| 432 |
],
|
| 433 |
outputs=[audio_output, stats],
|
|
|
|
| 440 |
generate.click(
|
| 441 |
synthesize,
|
| 442 |
inputs=[
|
| 443 |
+
text, language, speaker, emotion, style_energy,
|
| 444 |
audio_temperature, audio_top_p, audio_top_k, audio_repetition_penalty,
|
| 445 |
],
|
| 446 |
outputs=[audio_output, stats],
|
|
|
|
| 449 |
|
| 450 |
if __name__ == "__main__":
|
| 451 |
demo.queue(default_concurrency_limit=1, max_size=20)
|
| 452 |
+
demo.launch(mcp_server=True, show_error=False)
|