Update app.py
Browse files
app.py
CHANGED
@@ -1,44 +1,59 @@
|
|
1 |
import gradio as gr
|
2 |
-
import torch
|
3 |
import torchaudio
|
4 |
from tortoise.api import TextToSpeech
|
5 |
from tortoise.utils.audio import load_voice
|
6 |
|
7 |
-
# Initialize
|
8 |
tts = TextToSpeech()
|
9 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
# Function to generate speech
|
11 |
def generate_speech(text, voice):
|
12 |
-
# Load the selected voice
|
13 |
-
voice_samples, conditioning_latents = load_voice(voice)
|
14 |
-
|
15 |
-
# Generate speech
|
16 |
-
with torch.no_grad():
|
17 |
-
gen = tts.tts_with_preset(text, voice_samples=voice_samples, conditioning_latents=conditioning_latents, preset="fast")
|
18 |
-
torchaudio.save("output.wav", gen.squeeze(0).cpu(), 24000)
|
19 |
-
|
20 |
-
return "output.wav"
|
21 |
-
|
22 |
-
# Gradio interface
|
23 |
-
def tts_app(text, voice):
|
24 |
if not text:
|
25 |
return "Please enter some text."
|
26 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
27 |
|
28 |
-
# Gradio
|
29 |
with gr.Blocks(
|
30 |
theme=gr.themes.Soft(primary_hue="teal", secondary_hue="pink"),
|
31 |
css=".gradio-container {background: linear-gradient(135deg, #f5f7fa, #c3cfe2);} "
|
32 |
"button {background: linear-gradient(135deg, #6a11cb, #2575fc); color: white; border: none; padding: 10px 20px; border-radius: 5px;} "
|
33 |
"button:hover {background: linear-gradient(135deg, #2575fc, #6a11cb);} "
|
34 |
) as demo:
|
|
|
35 |
gr.Markdown(
|
36 |
"""
|
37 |
-
# 🎤 **
|
38 |
-
**
|
39 |
"""
|
40 |
)
|
41 |
|
|
|
42 |
with gr.Row():
|
43 |
text_input = gr.Textbox(
|
44 |
lines=5,
|
@@ -46,17 +61,20 @@ with gr.Blocks(
|
|
46 |
placeholder="Type your text here..."
|
47 |
)
|
48 |
voice_input = gr.Dropdown(
|
49 |
-
choices=
|
50 |
label="🗣️ **Select Voice**",
|
51 |
-
value="
|
52 |
)
|
53 |
|
|
|
54 |
output_audio = gr.Audio(label="🎧 **Generated Speech**", type="filepath")
|
55 |
|
|
|
56 |
submit_button = gr.Button("✨ **Generate Speech**")
|
57 |
|
|
|
58 |
submit_button.click(
|
59 |
-
|
60 |
inputs=[text_input, voice_input],
|
61 |
outputs=output_audio
|
62 |
)
|
|
|
1 |
import gradio as gr
|
|
|
2 |
import torchaudio
|
3 |
from tortoise.api import TextToSpeech
|
4 |
from tortoise.utils.audio import load_voice
|
5 |
|
6 |
+
# Initialize TorToiSe
|
7 |
tts = TextToSpeech()
|
8 |
|
9 |
+
# List of available voices
|
10 |
+
AVAILABLE_VOICES = [
|
11 |
+
"angie", "daniel", "deniro", "emma", "freeman", "geralt", "halle",
|
12 |
+
"jlaw", "lj", "myself", "pat", "snakes", "tom", "train_atkins",
|
13 |
+
"train_dotrice", "train_kennard", "weaver", "william"
|
14 |
+
]
|
15 |
+
|
16 |
# Function to generate speech
|
17 |
def generate_speech(text, voice):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
18 |
if not text:
|
19 |
return "Please enter some text."
|
20 |
+
|
21 |
+
try:
|
22 |
+
# Load the selected voice
|
23 |
+
voice_samples, conditioning_latents = load_voice(voice)
|
24 |
+
|
25 |
+
# Generate speech
|
26 |
+
gen = tts.tts_with_preset(
|
27 |
+
text,
|
28 |
+
voice_samples=voice_samples,
|
29 |
+
conditioning_latents=conditioning_latents,
|
30 |
+
preset="fast"
|
31 |
+
)
|
32 |
+
|
33 |
+
# Save the output
|
34 |
+
output_file = "output.wav"
|
35 |
+
torchaudio.save(output_file, gen.squeeze(0).cpu(), 24000)
|
36 |
+
|
37 |
+
return output_file
|
38 |
+
except Exception as e:
|
39 |
+
return f"Error generating speech: {str(e)}"
|
40 |
|
41 |
+
# Gradio interface
|
42 |
with gr.Blocks(
|
43 |
theme=gr.themes.Soft(primary_hue="teal", secondary_hue="pink"),
|
44 |
css=".gradio-container {background: linear-gradient(135deg, #f5f7fa, #c3cfe2);} "
|
45 |
"button {background: linear-gradient(135deg, #6a11cb, #2575fc); color: white; border: none; padding: 10px 20px; border-radius: 5px;} "
|
46 |
"button:hover {background: linear-gradient(135deg, #2575fc, #6a11cb);} "
|
47 |
) as demo:
|
48 |
+
# Title and description
|
49 |
gr.Markdown(
|
50 |
"""
|
51 |
+
# 🎤 **TorToiSe Text-to-Speech**
|
52 |
+
**Generate natural-sounding speech in multiple voices!**
|
53 |
"""
|
54 |
)
|
55 |
|
56 |
+
# Inputs
|
57 |
with gr.Row():
|
58 |
text_input = gr.Textbox(
|
59 |
lines=5,
|
|
|
61 |
placeholder="Type your text here..."
|
62 |
)
|
63 |
voice_input = gr.Dropdown(
|
64 |
+
choices=AVAILABLE_VOICES,
|
65 |
label="🗣️ **Select Voice**",
|
66 |
+
value="emma" # Default voice
|
67 |
)
|
68 |
|
69 |
+
# Output
|
70 |
output_audio = gr.Audio(label="🎧 **Generated Speech**", type="filepath")
|
71 |
|
72 |
+
# Submit button
|
73 |
submit_button = gr.Button("✨ **Generate Speech**")
|
74 |
|
75 |
+
# Link button to function
|
76 |
submit_button.click(
|
77 |
+
generate_speech,
|
78 |
inputs=[text_input, voice_input],
|
79 |
outputs=output_audio
|
80 |
)
|