Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import torch
|
3 |
+
import numpy as np
|
4 |
+
import scipy
|
5 |
+
from transformers import VitsModel, AutoTokenizer
|
6 |
+
|
7 |
+
# Load the model and tokenizer
|
8 |
+
model = VitsModel.from_pretrained("kakao-enterprise/vits-ljs")
|
9 |
+
tokenizer = AutoTokenizer.from_pretrained("kakao-enterprise/vits-ljs")
|
10 |
+
|
11 |
+
def text_to_speech(text):
|
12 |
+
# Tokenize the text
|
13 |
+
inputs = tokenizer(text, return_tensors="pt")
|
14 |
+
|
15 |
+
# Generate audio
|
16 |
+
with torch.no_grad():
|
17 |
+
output = model(**inputs).waveform
|
18 |
+
|
19 |
+
# Convert to numpy array and save as WAV file
|
20 |
+
audio_array = output.cpu().numpy().squeeze()
|
21 |
+
audio_array /= 1.414
|
22 |
+
audio_array *= 32767
|
23 |
+
audio_array = audio_array.astype(np.int16)
|
24 |
+
|
25 |
+
# Save to WAV file
|
26 |
+
output_file = "output/output.wav"
|
27 |
+
scipy.io.wavfile.write(output_file, rate=model.config.sampling_rate, data=audio_array)
|
28 |
+
|
29 |
+
# Return the path to the WAV file
|
30 |
+
return output_file
|
31 |
+
|
32 |
+
demo = gr.Interface(
|
33 |
+
text_to_speech,
|
34 |
+
gr.Textbox(label="Text to narrate"),
|
35 |
+
gr.Audio(label="Narrated audio"),
|
36 |
+
title="Text-to-Speech",
|
37 |
+
description="Enter text to generate audio narration",
|
38 |
+
)
|
39 |
+
|
40 |
+
demo.launch()
|