Spaces:
Runtime error
Runtime error
Create main.py
Browse files
main.py
ADDED
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from fastapi import FastAPI, Query, HTTPException
|
2 |
+
from fastapi.responses import StreamingResponse
|
3 |
+
from TTS.api import TTS
|
4 |
+
import os
|
5 |
+
from io import BytesIO
|
6 |
+
from typing import Generator
|
7 |
+
|
8 |
+
app = FastAPI()
|
9 |
+
|
10 |
+
# Initialize the TTS model
|
11 |
+
tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2", gpu=False) # Set gpu=True if you have GPU support
|
12 |
+
|
13 |
+
# Function to split text into chunks
|
14 |
+
def split_text(text: str, words_per_chunk: int = 20):
|
15 |
+
words = text.split()
|
16 |
+
return [' '.join(words[i:i + words_per_chunk]) for i in range(0, len(words), words_per_chunk)]
|
17 |
+
|
18 |
+
# Function to generate audio chunks
|
19 |
+
def generate_audio_chunks(
|
20 |
+
text: str, speaker_wav: str, language: str, chunk_size: int = 20
|
21 |
+
) -> Generator[bytes, None, None]:
|
22 |
+
if tts.is_multi_lingual and not language:
|
23 |
+
raise ValueError("Language must be specified for multi-lingual models.")
|
24 |
+
|
25 |
+
text_chunks = split_text(text, chunk_size)
|
26 |
+
|
27 |
+
for idx, chunk in enumerate(text_chunks):
|
28 |
+
# Generate audio for each chunk and yield as bytes
|
29 |
+
audio_buffer = BytesIO()
|
30 |
+
tts.tts_to_file(
|
31 |
+
text=chunk,
|
32 |
+
file_path=audio_buffer,
|
33 |
+
speaker_wav=speaker_wav,
|
34 |
+
language=language
|
35 |
+
)
|
36 |
+
audio_buffer.seek(0)
|
37 |
+
yield audio_buffer.read()
|
38 |
+
|
39 |
+
@app.post("/generate-audio/")
|
40 |
+
async def generate_audio(
|
41 |
+
text: str = Query(..., description="The input text to convert to speech."),
|
42 |
+
language: str = Query("en", description="Language code for TTS (e.g., 'en' for English)."),
|
43 |
+
speaker_wav: str = Query(..., description="Path to the WAV file for voice cloning.")
|
44 |
+
):
|
45 |
+
if not os.path.exists(speaker_wav):
|
46 |
+
raise HTTPException(status_code=400, detail="Speaker WAV file not found.")
|
47 |
+
|
48 |
+
# StreamingResponse to stream audio chunks
|
49 |
+
def audio_stream():
|
50 |
+
try:
|
51 |
+
for audio_chunk in generate_audio_chunks(text=text, speaker_wav=speaker_wav, language=language):
|
52 |
+
yield audio_chunk
|
53 |
+
except Exception as e:
|
54 |
+
raise HTTPException(status_code=500, detail=str(e))
|
55 |
+
|
56 |
+
return StreamingResponse(audio_stream(), media_type="audio/wav")
|