Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -155,8 +155,15 @@ async def generate_story_audios(story: StoryCreationDTO, base_output: str):
|
|
| 155 |
# --- Chapter title audio ---
|
| 156 |
prosody_file_title = await download_file_from_url(chapter.title.prosodyReference)
|
| 157 |
title_save_path = chapter_dir / "title.wav"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 158 |
title_generated_audio_path = inference_by_model(
|
| 159 |
-
text=
|
| 160 |
audio_file=prosody_file_title,
|
| 161 |
save_path=title_save_path
|
| 162 |
)
|
|
@@ -171,8 +178,13 @@ async def generate_story_audios(story: StoryCreationDTO, base_output: str):
|
|
| 171 |
# Download the prosody reference audio from Supabase
|
| 172 |
prosody_file = await download_file_from_url(sentence.prosodyReference)
|
| 173 |
sentence_save_path = scene_dir / f"{sentence.sentenceId}.wav"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 174 |
sentence_generated_audio_path = inference_by_model(
|
| 175 |
-
text=
|
| 176 |
audio_file=prosody_file,
|
| 177 |
save_path=sentence_save_path
|
| 178 |
)
|
|
@@ -298,6 +310,35 @@ def audio_to_base64(audio_path: str) -> (str, float):
|
|
| 298 |
audio_b64 = base64.b64encode(audio_bytes).decode("utf-8")
|
| 299 |
return audio_b64, duration
|
| 300 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 301 |
|
| 302 |
@app.post("/tts/")
|
| 303 |
async def process_story(story: StoryCreationDTO):
|
|
@@ -329,35 +370,7 @@ async def process_story(story: StoryCreationDTO):
|
|
| 329 |
|
| 330 |
return response
|
| 331 |
|
| 332 |
-
#---------------------------
|
| 333 |
-
|
| 334 |
-
# Map Intensity numbers to tag strings
|
| 335 |
-
intensity_map = {
|
| 336 |
-
1: "low",
|
| 337 |
-
2: "mid",
|
| 338 |
-
3: "high"
|
| 339 |
-
}
|
| 340 |
-
|
| 341 |
-
# Map Emotion enum names to lowercase tag strings
|
| 342 |
-
emotion_map = {
|
| 343 |
-
"HAPPINESS": "happiness",
|
| 344 |
-
"SADNESS": "sadness",
|
| 345 |
-
"FEAR": "fear",
|
| 346 |
-
"ANGER": "anger",
|
| 347 |
-
"SURPRISE": "surprise",
|
| 348 |
-
"WHISPER": "whisper",
|
| 349 |
-
"NARRATION": "narration"
|
| 350 |
-
}
|
| 351 |
-
|
| 352 |
-
def generate_tagged_text(text: str, emotion_enum: str, intensity_enum: int) -> str:
|
| 353 |
-
"""
|
| 354 |
-
Convert enums to <emo_x> <int_y> format and concatenate with text
|
| 355 |
-
"""
|
| 356 |
-
emo_tag = f"<emo_{emotion_map[emotion_enum]}>"
|
| 357 |
-
int_tag = f"<int_{intensity_map[intensity_enum]}>"
|
| 358 |
-
return f"{emo_tag} {int_tag} {text}"
|
| 359 |
-
|
| 360 |
-
#-----------------------------------------------------------
|
| 361 |
|
| 362 |
@app.post("/tts_test/")
|
| 363 |
async def tts_endpoint(
|
|
|
|
| 155 |
# --- Chapter title audio ---
|
| 156 |
prosody_file_title = await download_file_from_url(chapter.title.prosodyReference)
|
| 157 |
title_save_path = chapter_dir / "title.wav"
|
| 158 |
+
|
| 159 |
+
tagged_text_title = generate_tagged_text(
|
| 160 |
+
chapter.title.sentence,
|
| 161 |
+
chapter.title.emotion,
|
| 162 |
+
chapter.title.intensity
|
| 163 |
+
)
|
| 164 |
+
|
| 165 |
title_generated_audio_path = inference_by_model(
|
| 166 |
+
text=tagged_text_title,
|
| 167 |
audio_file=prosody_file_title,
|
| 168 |
save_path=title_save_path
|
| 169 |
)
|
|
|
|
| 178 |
# Download the prosody reference audio from Supabase
|
| 179 |
prosody_file = await download_file_from_url(sentence.prosodyReference)
|
| 180 |
sentence_save_path = scene_dir / f"{sentence.sentenceId}.wav"
|
| 181 |
+
tagged_text = generate_tagged_text(
|
| 182 |
+
sentence.sentence,
|
| 183 |
+
sentence.emotion,
|
| 184 |
+
sentence.intensity
|
| 185 |
+
)
|
| 186 |
sentence_generated_audio_path = inference_by_model(
|
| 187 |
+
text=tagged_text,
|
| 188 |
audio_file=prosody_file,
|
| 189 |
save_path=sentence_save_path
|
| 190 |
)
|
|
|
|
| 310 |
audio_b64 = base64.b64encode(audio_bytes).decode("utf-8")
|
| 311 |
return audio_b64, duration
|
| 312 |
|
| 313 |
+
#---------------------------concatenate text with tags ---------------------------
|
| 314 |
+
|
| 315 |
+
# Map Intensity numbers to tag strings
|
| 316 |
+
intensity_map = {
|
| 317 |
+
1: "low",
|
| 318 |
+
2: "mid",
|
| 319 |
+
3: "high"
|
| 320 |
+
}
|
| 321 |
+
|
| 322 |
+
# Map Emotion enum names to lowercase tag strings
|
| 323 |
+
emotion_map = {
|
| 324 |
+
"HAPPINESS": "happiness",
|
| 325 |
+
"SADNESS": "sadness",
|
| 326 |
+
"FEAR": "fear",
|
| 327 |
+
"ANGER": "anger",
|
| 328 |
+
"SURPRISE": "surprise",
|
| 329 |
+
"WHISPER": "whisper",
|
| 330 |
+
"NARRATION": "narration"
|
| 331 |
+
}
|
| 332 |
+
|
| 333 |
+
def generate_tagged_text(text: str, emotion_enum: str, intensity_enum: int) -> str:
|
| 334 |
+
"""
|
| 335 |
+
Convert enums to <emo_x> <int_y> format and concatenate with text
|
| 336 |
+
"""
|
| 337 |
+
emo_tag = f"<emo_{emotion_map[emotion_enum]}>"
|
| 338 |
+
int_tag = f"<int_{intensity_map[intensity_enum]}>"
|
| 339 |
+
return f"{emo_tag} {int_tag} {text}"
|
| 340 |
+
|
| 341 |
+
#-----------------------------------------------------------
|
| 342 |
|
| 343 |
@app.post("/tts/")
|
| 344 |
async def process_story(story: StoryCreationDTO):
|
|
|
|
| 370 |
|
| 371 |
return response
|
| 372 |
|
| 373 |
+
#----------------------------Test------------------------------------
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 374 |
|
| 375 |
@app.post("/tts_test/")
|
| 376 |
async def tts_endpoint(
|