akhaliq HF staff commited on
Commit
a67942c
1 Parent(s): 0d91fca

Create new file

Browse files
Files changed (1) hide show
  1. app.py +164 -0
app.py ADDED
@@ -0,0 +1,164 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import gradio as gr
2
+ import whisper
3
+
4
+ model = whisper.load_model("base")
5
+
6
+
7
+
8
+
9
+ def inference(audio):
10
+ result = model.transcribe(audio)
11
+ print(result["text"])
12
+ return result["text"]
13
+
14
+
15
+ title="Whisper"
16
+
17
+ description="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."
18
+
19
+ css = """
20
+ .gradio-container {
21
+ font-family: 'IBM Plex Sans', sans-serif;
22
+ }
23
+ .gr-button {
24
+ color: white;
25
+ border-color: black;
26
+ background: black;
27
+ }
28
+ input[type='range'] {
29
+ accent-color: black;
30
+ }
31
+ .dark input[type='range'] {
32
+ accent-color: #dfdfdf;
33
+ }
34
+ .container {
35
+ max-width: 730px;
36
+ margin: auto;
37
+ padding-top: 1.5rem;
38
+ }
39
+ #gallery {
40
+ min-height: 22rem;
41
+ margin-bottom: 15px;
42
+ margin-left: auto;
43
+ margin-right: auto;
44
+ border-bottom-right-radius: .5rem !important;
45
+ border-bottom-left-radius: .5rem !important;
46
+ }
47
+ #gallery>div>.h-full {
48
+ min-height: 20rem;
49
+ }
50
+ .details:hover {
51
+ text-decoration: underline;
52
+ }
53
+ .gr-button {
54
+ white-space: nowrap;
55
+ }
56
+ .gr-button:focus {
57
+ border-color: rgb(147 197 253 / var(--tw-border-opacity));
58
+ outline: none;
59
+ box-shadow: var(--tw-ring-offset-shadow), var(--tw-ring-shadow), var(--tw-shadow, 0 0 #0000);
60
+ --tw-border-opacity: 1;
61
+ --tw-ring-offset-shadow: var(--tw-ring-inset) 0 0 0 var(--tw-ring-offset-width) var(--tw-ring-offset-color);
62
+ --tw-ring-shadow: var(--tw-ring-inset) 0 0 0 calc(3px var(--tw-ring-offset-width)) var(--tw-ring-color);
63
+ --tw-ring-color: rgb(191 219 254 / var(--tw-ring-opacity));
64
+ --tw-ring-opacity: .5;
65
+ }
66
+ .footer {
67
+ margin-bottom: 45px;
68
+ margin-top: 35px;
69
+ text-align: center;
70
+ border-bottom: 1px solid #e5e5e5;
71
+ }
72
+ .footer>p {
73
+ font-size: .8rem;
74
+ display: inline-block;
75
+ padding: 0 10px;
76
+ transform: translateY(10px);
77
+ background: white;
78
+ }
79
+ .dark .footer {
80
+ border-color: #303030;
81
+ }
82
+ .dark .footer>p {
83
+ background: #0b0f19;
84
+ }
85
+ .prompt h4{
86
+ margin: 1.25em 0 .25em 0;
87
+ font-weight: bold;
88
+ font-size: 115%;
89
+ }
90
+ """
91
+
92
+ block = gr.Blocks(css=css)
93
+
94
+
95
+
96
+ with block:
97
+ gr.HTML(
98
+ """
99
+ <div style="text-align: center; max-width: 650px; margin: 0 auto;">
100
+ <div
101
+ style="
102
+ display: inline-flex;
103
+ gap: 0.8rem;
104
+ font-size: 1.75rem;
105
+ margin-bottom: 10px;
106
+ margin-left: 220px;
107
+ justify-content: center;
108
+ "
109
+ >
110
+ <a href="https://github.com/PaddlePaddle/PaddleHub"><img src="https://user-images.githubusercontent.com/22424850/187387422-f6c9ccab-7fda-416e-a24d-7d6084c46f67.jpg" alt="Paddlehub" width="40%"></a>
111
+ </div>
112
+ <div
113
+ style="
114
+ display: inline-flex;
115
+ align-items: center;
116
+ gap: 0.8rem;
117
+ font-size: 1.75rem;
118
+ margin-bottom: 10px;
119
+ justify-content: center;
120
+ ">
121
+ <a href="https://github.com/PaddlePaddle/PaddleHub"><h1 style="font-weight: 900; margin-bottom: 7px;">
122
+ ERNIE-ViLG Demo
123
+ </h1></a>
124
+ </div>
125
+ <p style="margin-bottom: 10px; font-size: 94%">
126
+ ERNIE-ViLG is a state-of-the-art text-to-image model that generates
127
+ images from Chinese text.
128
+ </p>
129
+ <a href="https://github.com/PaddlePaddle/PaddleHub"><img src="https://user-images.githubusercontent.com/22424850/188184795-98605a22-9af2-4106-827b-e58548f8892f.png" alt="star Paddlehub" width="100%"></a>
130
+ </div>
131
+ """
132
+ )
133
+ with gr.Group():
134
+ with gr.Box():
135
+ with gr.Row().style(mobile_collapse=False, equal_height=True):
136
+ audio = gr.Audio(
137
+ label="Input Audio",
138
+ show_label=False,
139
+ ).style(
140
+ border=(True, False, True, True),
141
+ rounded=(True, False, False, True),
142
+ container=False,
143
+ )
144
+
145
+ btn = gr.Button("Transcribe").style(
146
+ margin=False,
147
+ rounded=(False, True, True, False),
148
+ )
149
+ text = gr.Textbox(
150
+ ).style(height="auto")
151
+
152
+
153
+
154
+
155
+ btn.click(inference, inputs=[audio], outputs=[text])
156
+
157
+ gr.HTML('''
158
+ <div class="footer">
159
+ <p>Model by <a href="https://github.com/openai/whisper" style="text-decoration: underline;" target="_blank">OpenAI</a> and <a href="https://wenxin.baidu.com" style="text-decoration: underline;" target="_blank">文心大模型</a> - Gradio Demo by 🤗 Hugging Face
160
+ </p>
161
+ </div>
162
+ ''')
163
+
164
+ block.launch()