Spaces:
Runtime error
Runtime error
Update main.py
Browse files
main.py
CHANGED
|
@@ -22,22 +22,7 @@ def split_text(text: str, words_per_chunk: int = 20):
|
|
| 22 |
words = text.split()
|
| 23 |
return [' '.join(words[i:i + words_per_chunk]) for i in range(0, len(words), words_per_chunk)]
|
| 24 |
|
| 25 |
-
|
| 26 |
-
def generate_audio_chunks(text: str, language: str, chunk_size: int = 20) -> Generator[bytes, None, None]:
|
| 27 |
-
if tts.is_multi_lingual and not language:
|
| 28 |
-
raise ValueError("Language must be specified for multi-lingual models.")
|
| 29 |
-
|
| 30 |
-
text_chunks = split_text(text, chunk_size)
|
| 31 |
-
|
| 32 |
-
for idx, chunk in enumerate(text_chunks):
|
| 33 |
-
# Generate audio for each chunk and yield as bytes
|
| 34 |
-
tts.tts_to_file(
|
| 35 |
-
text=chunk,
|
| 36 |
-
file_path=f"out_{idx}.wav",
|
| 37 |
-
speaker_wav=FIXED_SPEAKER_WAV,
|
| 38 |
-
language=language
|
| 39 |
-
)
|
| 40 |
-
yield f"out_{idx}.wav"
|
| 41 |
|
| 42 |
@app.post("/generate-audio/")
|
| 43 |
async def generate_audio(
|
|
@@ -49,10 +34,19 @@ async def generate_audio(
|
|
| 49 |
|
| 50 |
# StreamingResponse to stream audio chunks
|
| 51 |
def audio_stream():
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 58 |
return StreamingResponse(audio_stream(), media_type="audio/wav")
|
|
|
|
| 22 |
words = text.split()
|
| 23 |
return [' '.join(words[i:i + words_per_chunk]) for i in range(0, len(words), words_per_chunk)]
|
| 24 |
|
| 25 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
|
| 27 |
@app.post("/generate-audio/")
|
| 28 |
async def generate_audio(
|
|
|
|
| 34 |
|
| 35 |
# StreamingResponse to stream audio chunks
|
| 36 |
def audio_stream():
|
| 37 |
+
if tts.is_multi_lingual and not language:
|
| 38 |
+
raise ValueError("Language must be specified for multi-lingual models.")
|
| 39 |
+
|
| 40 |
+
text_chunks = split_text(text, 20)
|
| 41 |
+
|
| 42 |
+
for idx, chunk in enumerate(text_chunks):
|
| 43 |
+
# Generate audio for each chunk and yield as bytes
|
| 44 |
+
tts.tts_to_file(
|
| 45 |
+
text=chunk,
|
| 46 |
+
file_path=f"out_{idx}.wav",
|
| 47 |
+
speaker_wav=FIXED_SPEAKER_WAV,
|
| 48 |
+
language=language
|
| 49 |
+
)
|
| 50 |
+
yield f"out_{idx}.wav"
|
| 51 |
+
|
| 52 |
return StreamingResponse(audio_stream(), media_type="audio/wav")
|