Spaces:
Sleeping
Sleeping
Delete app.py
Browse files
app.py
DELETED
@@ -1,125 +0,0 @@
|
|
1 |
-
import os
|
2 |
-
import gradio as gr
|
3 |
-
from src.gradio_demo import SadTalker
|
4 |
-
|
5 |
-
try:
|
6 |
-
import webui # in webui
|
7 |
-
IN_WEBUi = True
|
8 |
-
except ImportError:
|
9 |
-
IN_WEBUi = False
|
10 |
-
|
11 |
-
def toggle_audio_file(choice):
|
12 |
-
if not choice:
|
13 |
-
return gr.update("visible", True), gr.update("visible", False)
|
14 |
-
else:
|
15 |
-
return gr.update("visible", False), gr.update("visible", True)
|
16 |
-
|
17 |
-
def ref_video_fn(path_of_ref_video):
|
18 |
-
if path_of_ref_video:
|
19 |
-
return gr.update("value", True)
|
20 |
-
else:
|
21 |
-
return gr.update("value", False)
|
22 |
-
|
23 |
-
def sadtalker_demo(checkpoint_path="checkpoints", config_path="src/config", warpfn=None):
|
24 |
-
sad_talker = SadTalker(checkpoint_path, config_path, lazy_load=True)
|
25 |
-
|
26 |
-
with gr.Blocks(analytics_enabled=False) as sadtalker_interface:
|
27 |
-
gr.markdown("<div align='center'> <h2> 😭 SadTalker: Learning Realistic 3D Motion Coefficients for Stylized Audio-Driven Single Image Talking Face Animation (CVPR 2023) </span> \
|
28 |
-
<a style='font-size:18px;color: #efefef' href='https://arxiv.org/abs/2211.12194'>Arxiv</a> \
|
29 |
-
<a style='font-size:18px;color: #efefef' href='https://sadtalker.github.io'>Homepage</a> \
|
30 |
-
<a style='font-size:18px;color: #efefef' href='https://github.com/Winfredy/SadTalker'> Github </div>")
|
31 |
-
|
32 |
-
with gr.Row(style="equal_height: False"):
|
33 |
-
with gr.Column(variant="panel"):
|
34 |
-
with gr.Tabs(elem_id="sadtalker_source_image"):
|
35 |
-
with gr.TabItem("Upload image"):
|
36 |
-
with gr.Row():
|
37 |
-
source_image = gr.Image(label="Source image", source="upload", type="filepath", elem_id="img2img_image", style="width: 512px")
|
38 |
-
|
39 |
-
with gr.Tabs(elem_id="sadtalker_driven_audio"):
|
40 |
-
with gr.TabItem("Upload OR TTS"):
|
41 |
-
with gr.Column(variant="panel"):
|
42 |
-
driven_audio = gr.Audio(label="Input audio", source="upload", type="filepath")
|
43 |
-
|
44 |
-
if sys.platform != "win32" and not IN_WEBUi:
|
45 |
-
from src.utils.text2speech import TTSTalker
|
46 |
-
|
47 |
-
tts_talker = TTSTalker()
|
48 |
-
with gr.Column(variant="panel"):
|
49 |
-
input_text = gr.Textbox(
|
50 |
-
label="Generating audio from text",
|
51 |
-
lines=5,
|
52 |
-
placeholder="Enter text for audio generation using @Coqui.ai TTS.",
|
53 |
-
)
|
54 |
-
tts = gr.Button("Generate audio", elem_id="sadtalker_audio_generate", variant="primary")
|
55 |
-
tts.click(tts_talker.test, [input_text], [driven_audio])
|
56 |
-
|
57 |
-
with gr.Column(variant="panel"):
|
58 |
-
with gr.Tabs(elem_id="sadtalker_checkbox"):
|
59 |
-
with gr.TabItem("Settings"):
|
60 |
-
gr.markdown(
|
61 |
-
"Need help? Visit our [best practice page](https://github.com/OpenTalker/SadTalker/blob/main/docs/best_practice.md) for details."
|
62 |
-
)
|
63 |
-
with gr.Column(variant="panel"):
|
64 |
-
# width = gr.Slider(
|
65 |
-
# minimum=64,
|
66 |
-
# elem_id="img2img_width",
|
67 |
-
# maximum=2048,
|
68 |
-
# step=8,
|
69 |
-
# label="Manually Crop Width",
|
70 |
-
# value=512,
|
71 |
-
# ) # img2img_width
|
72 |
-
# height = gr.Slider(
|
73 |
-
# minimum=64,
|
74 |
-
# elem_id="img2img_height",
|
75 |
-
# maximum=2048,
|
76 |
-
# step=8,
|
77 |
-
# label="Manually Crop Height",
|
78 |
-
# value=512,
|
79 |
-
# ) # img2img_width
|
80 |
-
pose_style = gr.Slider(
|
81 |
-
minimum=0, maximum=46, step=1, label="Pose style", value=0
|
82 |
-
)
|
83 |
-
size_of_image = gr.Radio(
|
84 |
-
[256, 512],
|
85 |
-
value=256,
|
86 |
-
label="face model resolution",
|
87 |
-
info="Use 256/512 model?",
|
88 |
-
)
|
89 |
-
preprocess_type = gr.Radio(
|
90 |
-
["crop", "resize", "full", "extcrop", "extfull"],
|
91 |
-
value="crop",
|
92 |
-
label="preprocess",
|
93 |
-
info="How to handle input image?",
|
94 |
-
)
|
95 |
-
is_still_mode = gr.Checkbox(
|
96 |
-
label="Still Mode (fewer head motion, works with preprocess `full`)"
|
97 |
-
)
|
98 |
-
batch_size = gr.Slider(
|
99 |
-
label="batch size in generation", step=1, maximum=10, value=2
|
100 |
-
)
|
101 |
-
enhancer = gr.Checkbox(label="GFPGAN as Face enhancer")
|
102 |
-
submit = gr.Button("Generate", elem_id="sadtalker_generate", variant="primary")
|
103 |
-
|
104 |
-
with gr.Tabs(elem_id="sadtalker_genearted"):
|
105 |
-
gen_video = gr.Video(label="Generated video", format="mp4", style="width: 256px")
|
106 |
-
|
107 |
-
if warpfn:
|
108 |
-
submit.click(
|
109 |
-
warpfn(sad_talker.test),
|
110 |
-
[source_image, driven_audio, preprocess_type, is_still_mode, enhancer, batch_size, size_of_image, pose_style],
|
111 |
-
[gen_video],
|
112 |
-
)
|
113 |
-
else:
|
114 |
-
submit.click(
|
115 |
-
sad_talker.test,
|
116 |
-
[source_image, driven_audio, preprocess_type, is_still_mode, enhancer, batch_size, size_of_image, pose_style],
|
117 |
-
[gen_video],
|
118 |
-
)
|
119 |
-
|
120 |
-
return sadtalker_interface
|
121 |
-
|
122 |
-
if __name__ == "__main__":
|
123 |
-
demo = sadtalker_demo()
|
124 |
-
demo.queue()
|
125 |
-
demo.launch()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|