Commit
•
2e0561d
1
Parent(s):
fc7482c
Update app.py
Browse files
app.py
CHANGED
@@ -18,144 +18,13 @@ def inference(audio):
|
|
18 |
result = whisper.decode(model, mel, options)
|
19 |
|
20 |
print(result.text)
|
21 |
-
return result.text
|
22 |
|
23 |
|
24 |
-
|
25 |
-
|
26 |
-
css = """
|
27 |
-
.gradio-container {
|
28 |
-
font-family: 'IBM Plex Sans', sans-serif;
|
29 |
-
}
|
30 |
-
.gr-button {
|
31 |
-
color: white;
|
32 |
-
border-color: black;
|
33 |
-
background: black;
|
34 |
-
}
|
35 |
-
input[type='range'] {
|
36 |
-
accent-color: black;
|
37 |
-
}
|
38 |
-
.dark input[type='range'] {
|
39 |
-
accent-color: #dfdfdf;
|
40 |
-
}
|
41 |
-
.container {
|
42 |
-
max-width: 730px;
|
43 |
-
margin: auto;
|
44 |
-
padding-top: 1.5rem;
|
45 |
-
}
|
46 |
-
|
47 |
-
.details:hover {
|
48 |
-
text-decoration: underline;
|
49 |
-
}
|
50 |
-
.gr-button {
|
51 |
-
white-space: nowrap;
|
52 |
-
}
|
53 |
-
.gr-button:focus {
|
54 |
-
border-color: rgb(147 197 253 / var(--tw-border-opacity));
|
55 |
-
outline: none;
|
56 |
-
box-shadow: var(--tw-ring-offset-shadow), var(--tw-ring-shadow), var(--tw-shadow, 0 0 #0000);
|
57 |
-
--tw-border-opacity: 1;
|
58 |
-
--tw-ring-offset-shadow: var(--tw-ring-inset) 0 0 0 var(--tw-ring-offset-width) var(--tw-ring-offset-color);
|
59 |
-
--tw-ring-shadow: var(--tw-ring-inset) 0 0 0 calc(3px var(--tw-ring-offset-width)) var(--tw-ring-color);
|
60 |
-
--tw-ring-color: rgb(191 219 254 / var(--tw-ring-opacity));
|
61 |
-
--tw-ring-opacity: .5;
|
62 |
-
}
|
63 |
-
.footer {
|
64 |
-
margin-bottom: 45px;
|
65 |
-
margin-top: 35px;
|
66 |
-
text-align: center;
|
67 |
-
border-bottom: 1px solid #e5e5e5;
|
68 |
-
}
|
69 |
-
.footer>p {
|
70 |
-
font-size: .8rem;
|
71 |
-
display: inline-block;
|
72 |
-
padding: 0 10px;
|
73 |
-
transform: translateY(10px);
|
74 |
-
background: white;
|
75 |
-
}
|
76 |
-
.dark .footer {
|
77 |
-
border-color: #303030;
|
78 |
-
}
|
79 |
-
.dark .footer>p {
|
80 |
-
background: #0b0f19;
|
81 |
-
}
|
82 |
-
.prompt h4{
|
83 |
-
margin: 1.25em 0 .25em 0;
|
84 |
-
font-weight: bold;
|
85 |
-
font-size: 115%;
|
86 |
-
}
|
87 |
-
.animate-spin {
|
88 |
-
animation: spin 1s linear infinite;
|
89 |
-
}
|
90 |
-
@keyframes spin {
|
91 |
-
from {
|
92 |
-
transform: rotate(0deg);
|
93 |
-
}
|
94 |
-
to {
|
95 |
-
transform: rotate(360deg);
|
96 |
-
}
|
97 |
-
}
|
98 |
-
"""
|
99 |
-
|
100 |
block = gr.Blocks(css=css)
|
101 |
|
102 |
|
103 |
with block:
|
104 |
-
gr.HTML(
|
105 |
-
"""
|
106 |
-
<div style="text-align: center; max-width: 650px; margin: 0 auto;">
|
107 |
-
<div
|
108 |
-
style="
|
109 |
-
display: inline-flex;
|
110 |
-
align-items: center;
|
111 |
-
gap: 0.8rem;
|
112 |
-
font-size: 1.75rem;
|
113 |
-
"
|
114 |
-
>
|
115 |
-
<svg
|
116 |
-
width="0.65em"
|
117 |
-
height="0.65em"
|
118 |
-
viewBox="0 0 115 115"
|
119 |
-
fill="none"
|
120 |
-
xmlns="http://www.w3.org/2000/svg"
|
121 |
-
>
|
122 |
-
<rect width="23" height="23" fill="white"></rect>
|
123 |
-
<rect y="69" width="23" height="23" fill="white"></rect>
|
124 |
-
<rect x="23" width="23" height="23" fill="#AEAEAE"></rect>
|
125 |
-
<rect x="23" y="69" width="23" height="23" fill="#AEAEAE"></rect>
|
126 |
-
<rect x="46" width="23" height="23" fill="white"></rect>
|
127 |
-
<rect x="46" y="69" width="23" height="23" fill="white"></rect>
|
128 |
-
<rect x="69" width="23" height="23" fill="black"></rect>
|
129 |
-
<rect x="69" y="69" width="23" height="23" fill="black"></rect>
|
130 |
-
<rect x="92" width="23" height="23" fill="#D9D9D9"></rect>
|
131 |
-
<rect x="92" y="69" width="23" height="23" fill="#AEAEAE"></rect>
|
132 |
-
<rect x="115" y="46" width="23" height="23" fill="white"></rect>
|
133 |
-
<rect x="115" y="115" width="23" height="23" fill="white"></rect>
|
134 |
-
<rect x="115" y="69" width="23" height="23" fill="#D9D9D9"></rect>
|
135 |
-
<rect x="92" y="46" width="23" height="23" fill="#AEAEAE"></rect>
|
136 |
-
<rect x="92" y="115" width="23" height="23" fill="#AEAEAE"></rect>
|
137 |
-
<rect x="92" y="69" width="23" height="23" fill="white"></rect>
|
138 |
-
<rect x="69" y="46" width="23" height="23" fill="white"></rect>
|
139 |
-
<rect x="69" y="115" width="23" height="23" fill="white"></rect>
|
140 |
-
<rect x="69" y="69" width="23" height="23" fill="#D9D9D9"></rect>
|
141 |
-
<rect x="46" y="46" width="23" height="23" fill="black"></rect>
|
142 |
-
<rect x="46" y="115" width="23" height="23" fill="black"></rect>
|
143 |
-
<rect x="46" y="69" width="23" height="23" fill="black"></rect>
|
144 |
-
<rect x="23" y="46" width="23" height="23" fill="#D9D9D9"></rect>
|
145 |
-
<rect x="23" y="115" width="23" height="23" fill="#AEAEAE"></rect>
|
146 |
-
<rect x="23" y="69" width="23" height="23" fill="black"></rect>
|
147 |
-
</svg>
|
148 |
-
<h1 style="font-weight: 900; margin-bottom: 7px;">
|
149 |
-
Whisper
|
150 |
-
</h1>
|
151 |
-
</div>
|
152 |
-
<p style="margin-bottom: 10px; font-size: 94%">
|
153 |
-
Whisper is a general-purpose speech recognition model. It is trained on a large dataset of diverse audio and is also a multi-task model that can perform multilingual speech recognition as well as speech translation and language identification. This demo cuts audio after around 30 secs.
|
154 |
-
</p>
|
155 |
-
<p>You can skip the queue by using google colab for the space: <a href="https://colab.research.google.com/drive/1WJ98KHgZxFGrHiMm4TyWZllSew_Af_ff?usp=sharing"><img data-canonical-src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab" src="https://camo.githubusercontent.com/84f0493939e0c4de4e6dbe113251b4bfb5353e57134ffd9fcab6b8714514d4d1/68747470733a2f2f636f6c61622e72657365617263682e676f6f676c652e636f6d2f6173736574732f636f6c61622d62616467652e737667"></a></p>
|
156 |
-
</div>
|
157 |
-
"""
|
158 |
-
)
|
159 |
with gr.Group():
|
160 |
with gr.Box():
|
161 |
with gr.Row().style(mobile_collapse=False, equal_height=True):
|
@@ -173,8 +42,7 @@ with block:
|
|
173 |
|
174 |
gr.HTML('''
|
175 |
<div class="footer">
|
176 |
-
<p>
|
177 |
-
</p>
|
178 |
</div>
|
179 |
''')
|
180 |
|
|
|
18 |
result = whisper.decode(model, mel, options)
|
19 |
|
20 |
print(result.text)
|
21 |
+
return result.text
|
22 |
|
23 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
block = gr.Blocks(css=css)
|
25 |
|
26 |
|
27 |
with block:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
28 |
with gr.Group():
|
29 |
with gr.Box():
|
30 |
with gr.Row().style(mobile_collapse=False, equal_height=True):
|
|
|
42 |
|
43 |
gr.HTML('''
|
44 |
<div class="footer">
|
45 |
+
<p>This space is used by VideoChain, the all-in-one API to generate video, audio and more.</p>
|
|
|
46 |
</div>
|
47 |
''')
|
48 |
|