Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -288,17 +288,37 @@ def get_all_phonetics_schemes(text):
|
|
| 288 |
|
| 289 |
|
| 290 |
# Tibetan TTS function
|
| 291 |
-
def run_task_tts(text):
|
| 292 |
# Always return: [audio_numpy, audio_filepath, text_output]
|
| 293 |
# 1) Generate speech via MMS-TTS
|
| 294 |
-
speech = tts_tibetan(text)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 295 |
# 2) Clip, cast, flatten for Gradio (browser playback expects float32 in [-1, 1])
|
| 296 |
audio = speech["audio"]
|
| 297 |
sr = int(speech["sampling_rate"])
|
| 298 |
audio = np.clip(audio.astype(np.float32), -1.0, 1.0).flatten()
|
| 299 |
-
|
|
|
|
| 300 |
tmpfile = tempfile.NamedTemporaryFile(delete=False, suffix=".wav")
|
| 301 |
sf.write(tmpfile.name, audio, sr, subtype="PCM_16")
|
|
|
|
| 302 |
# 4) Return both audio forms + a status message
|
| 303 |
return (sr, audio), tmpfile.name, "Tibetan audio generated successfully!"
|
| 304 |
|
|
@@ -567,15 +587,36 @@ def generate_tts_file(text: str) -> str:
|
|
| 567 |
_, file_path, _ = run_task_tts(text) # unpack tuple
|
| 568 |
return file_path
|
| 569 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 570 |
@api.post("/api/tts")
|
| 571 |
async def api_tts(request: gr.Request):
|
| 572 |
body = await request.json()
|
| 573 |
text = body.get("text", "")
|
| 574 |
|
| 575 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 576 |
return {"error": "No text provided"}
|
| 577 |
|
| 578 |
-
|
|
|
|
| 579 |
|
| 580 |
return FileResponse(
|
| 581 |
output_path,
|
|
@@ -584,6 +625,7 @@ async def api_tts(request: gr.Request):
|
|
| 584 |
)
|
| 585 |
|
| 586 |
|
|
|
|
| 587 |
#############################################
|
| 588 |
# 🔥 Attach your existing Gradio UI
|
| 589 |
#############################################
|
|
|
|
| 288 |
|
| 289 |
|
| 290 |
# Tibetan TTS function
|
| 291 |
+
#def run_task_tts(text):
|
| 292 |
# Always return: [audio_numpy, audio_filepath, text_output]
|
| 293 |
# 1) Generate speech via MMS-TTS
|
| 294 |
+
# speech = tts_tibetan(text)
|
| 295 |
+
# 2) Clip, cast, flatten for Gradio (browser playback expects float32 in [-1, 1])
|
| 296 |
+
# audio = speech["audio"]
|
| 297 |
+
# sr = int(speech["sampling_rate"])
|
| 298 |
+
# audio = np.clip(audio.astype(np.float32), -1.0, 1.0).flatten()
|
| 299 |
+
# 3) Write a WAV file for download/Flutter using PCM_16 to avoid pydub header errors
|
| 300 |
+
# tmpfile = tempfile.NamedTemporaryFile(delete=False, suffix=".wav")
|
| 301 |
+
# sf.write(tmpfile.name, audio, sr, subtype="PCM_16")
|
| 302 |
+
# 4) Return both audio forms + a status message
|
| 303 |
+
# return (sr, audio), tmpfile.name, "Tibetan audio generated successfully!"
|
| 304 |
+
|
| 305 |
+
def run_task_tts(text: str):
|
| 306 |
+
# Ensure input is a string
|
| 307 |
+
if not isinstance(text, str):
|
| 308 |
+
text = str(text)
|
| 309 |
+
|
| 310 |
+
# 1) Generate speech via MMS-TTS
|
| 311 |
+
speech = tts_tibetan(text) # pipeline expects plain string
|
| 312 |
+
|
| 313 |
# 2) Clip, cast, flatten for Gradio (browser playback expects float32 in [-1, 1])
|
| 314 |
audio = speech["audio"]
|
| 315 |
sr = int(speech["sampling_rate"])
|
| 316 |
audio = np.clip(audio.astype(np.float32), -1.0, 1.0).flatten()
|
| 317 |
+
|
| 318 |
+
# 3) Write a WAV file for download/Flutter using PCM_16
|
| 319 |
tmpfile = tempfile.NamedTemporaryFile(delete=False, suffix=".wav")
|
| 320 |
sf.write(tmpfile.name, audio, sr, subtype="PCM_16")
|
| 321 |
+
|
| 322 |
# 4) Return both audio forms + a status message
|
| 323 |
return (sr, audio), tmpfile.name, "Tibetan audio generated successfully!"
|
| 324 |
|
|
|
|
| 587 |
_, file_path, _ = run_task_tts(text) # unpack tuple
|
| 588 |
return file_path
|
| 589 |
|
| 590 |
+
#@api.post("/api/tts")
|
| 591 |
+
#async def api_tts(request: gr.Request):
|
| 592 |
+
# body = await request.json()
|
| 593 |
+
# text = body.get("text", "")
|
| 594 |
+
|
| 595 |
+
# if not text:
|
| 596 |
+
# return {"error": "No text provided"}
|
| 597 |
+
|
| 598 |
+
# output_path = generate_tts_file(text)
|
| 599 |
+
|
| 600 |
+
# return FileResponse(
|
| 601 |
+
# output_path,
|
| 602 |
+
# media_type="audio/wav",
|
| 603 |
+
# filename="tts.wav"
|
| 604 |
+
# )
|
| 605 |
+
|
| 606 |
@api.post("/api/tts")
|
| 607 |
async def api_tts(request: gr.Request):
|
| 608 |
body = await request.json()
|
| 609 |
text = body.get("text", "")
|
| 610 |
|
| 611 |
+
# ✅ Ensure text is always a string
|
| 612 |
+
if not isinstance(text, str):
|
| 613 |
+
text = str(text)
|
| 614 |
+
|
| 615 |
+
if not text.strip():
|
| 616 |
return {"error": "No text provided"}
|
| 617 |
|
| 618 |
+
# Call your wrapper
|
| 619 |
+
_, output_path, status = run_task_tts(text)
|
| 620 |
|
| 621 |
return FileResponse(
|
| 622 |
output_path,
|
|
|
|
| 625 |
)
|
| 626 |
|
| 627 |
|
| 628 |
+
|
| 629 |
#############################################
|
| 630 |
# 🔥 Attach your existing Gradio UI
|
| 631 |
#############################################
|