drewThomasson
commited on
Commit
•
4e243b0
1
Parent(s):
2a7ff02
Update app.py
Browse files
app.py
CHANGED
@@ -2,43 +2,80 @@ import json
|
|
2 |
import torch
|
3 |
from TTS.api import TTS
|
4 |
import gradio as gr
|
|
|
5 |
|
6 |
-
#
|
7 |
-
|
8 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
|
10 |
# Get device
|
11 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
12 |
|
13 |
# Function to generate audio
|
14 |
-
def generate_audio(text,
|
15 |
-
model_name = f"tts_models/{
|
16 |
tts = TTS(model_name=model_name, progress_bar=False).to(device)
|
17 |
output_path = "output_file.wav"
|
18 |
tts.tts_to_file(text=text, file_path=output_path)
|
19 |
return output_path
|
20 |
|
21 |
-
#
|
22 |
-
|
23 |
-
|
|
|
|
|
|
|
24 |
|
25 |
-
#
|
26 |
-
def
|
27 |
-
|
28 |
-
# Map selected language to its ISO code
|
29 |
-
iso_code = iso_codes[languages.index(language)]
|
30 |
-
return iso_code
|
31 |
|
|
|
|
|
32 |
with gr.Blocks() as demo:
|
33 |
with gr.Row():
|
34 |
-
|
|
|
|
|
|
|
|
|
|
|
35 |
language_dropdown = gr.Dropdown(
|
36 |
-
|
|
|
|
|
|
|
37 |
)
|
|
|
|
|
|
|
|
|
38 |
|
39 |
-
|
40 |
-
|
|
|
|
|
|
|
|
|
41 |
|
|
|
42 |
generate_button.click(
|
43 |
fn=lambda text, lang: generate_audio(text, update_iso_code(lang)),
|
44 |
inputs=[text_input, language_dropdown],
|
|
|
2 |
import torch
|
3 |
from TTS.api import TTS
|
4 |
import gradio as gr
|
5 |
+
from bs4 import BeautifulSoup
|
6 |
|
7 |
+
# Function to parse HTML and create a language dictionary
|
8 |
+
def parse_html_to_dict(file_path):
|
9 |
+
with open(file_path, 'r', encoding='utf-8') as file:
|
10 |
+
html_content = file.read()
|
11 |
+
|
12 |
+
soup = BeautifulSoup(html_content, 'html.parser')
|
13 |
+
language_dict = {}
|
14 |
+
for p in soup.find_all('p')[1:]:
|
15 |
+
parts = p.get_text().split()
|
16 |
+
if len(parts) >= 2:
|
17 |
+
code = parts[0].strip()
|
18 |
+
name = " ".join(parts[1:]).strip()
|
19 |
+
language_dict[name] = code
|
20 |
+
return language_dict
|
21 |
+
|
22 |
+
# Load the language dictionary from the HTML file
|
23 |
+
file_path = '1100_lan.htm' # Update this with your actual file path
|
24 |
+
language_dict = parse_html_to_dict(file_path)
|
25 |
+
languages = list(language_dict.keys())
|
26 |
+
iso_codes = list(language_dict.values())
|
27 |
|
28 |
# Get device
|
29 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
30 |
|
31 |
# Function to generate audio
|
32 |
+
def generate_audio(text, language_code):
|
33 |
+
model_name = f"tts_models/{language_code}/fairseq/vits"
|
34 |
tts = TTS(model_name=model_name, progress_bar=False).to(device)
|
35 |
output_path = "output_file.wav"
|
36 |
tts.tts_to_file(text=text, file_path=output_path)
|
37 |
return output_path
|
38 |
|
39 |
+
# Function to filter languages based on search term
|
40 |
+
def filter_languages(search_term):
|
41 |
+
if not search_term:
|
42 |
+
return gr.update(choices=languages, value=None)
|
43 |
+
filtered = [lang for lang in languages if search_term.lower() in lang.lower()]
|
44 |
+
return gr.update(choices=filtered, value=None if not filtered else filtered[0])
|
45 |
|
46 |
+
# Function to map selected language to ISO code
|
47 |
+
def update_iso_code(selected_language):
|
48 |
+
return language_dict.get(selected_language, "Unknown")
|
|
|
|
|
|
|
49 |
|
50 |
+
# Gradio interface
|
51 |
+
def run_interface():
|
52 |
with gr.Blocks() as demo:
|
53 |
with gr.Row():
|
54 |
+
search_input = gr.Textbox(
|
55 |
+
label="Search Language",
|
56 |
+
placeholder="Type to search...",
|
57 |
+
show_label=True
|
58 |
+
)
|
59 |
+
with gr.Row():
|
60 |
language_dropdown = gr.Dropdown(
|
61 |
+
choices=languages,
|
62 |
+
label="Select Language",
|
63 |
+
show_label=True,
|
64 |
+
interactive=True
|
65 |
)
|
66 |
+
with gr.Row():
|
67 |
+
text_input = gr.Textbox(label="Enter Text", lines=4, placeholder="Type your text here...")
|
68 |
+
generate_button = gr.Button("Generate Audio")
|
69 |
+
output_audio = gr.Audio(label="Generated Audio", type="filepath")
|
70 |
|
71 |
+
# Update dropdown choices when search input changes
|
72 |
+
search_input.change(
|
73 |
+
fn=filter_languages,
|
74 |
+
inputs=[search_input],
|
75 |
+
outputs=language_dropdown
|
76 |
+
)
|
77 |
|
78 |
+
# Generate audio when button is clicked
|
79 |
generate_button.click(
|
80 |
fn=lambda text, lang: generate_audio(text, update_iso_code(lang)),
|
81 |
inputs=[text_input, language_dropdown],
|