File size: 8,042 Bytes
2dbea9d
eb21a2f
58ffcfd
ce3ffca
296c11e
eb21a2f
 
bb9d46e
ce3ffca
 
25940ed
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2dbea9d
296c11e
c9cdb67
4ffbe02
5b8e16a
 
 
4ffbe02
eb21a2f
 
 
 
 
c9cdb67
 
296c11e
c9cdb67
 
 
 
 
 
 
 
 
eb21a2f
296c11e
 
c9cdb67
 
 
 
 
 
296c11e
c9cdb67
 
 
 
 
eb21a2f
4ffbe02
4edabe1
 
eb21a2f
3db7f09
 
25940ed
296c11e
 
 
eb21a2f
c9cdb67
 
 
296c11e
eb21a2f
 
 
 
 
 
296c11e
25940ed
 
 
 
 
 
58ffcfd
c9cdb67
296c11e
eb21a2f
25940ed
efa07b3
c9cdb67
eb21a2f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c9cdb67
 
 
 
 
 
 
 
 
 
 
25940ed
 
 
 
 
eb21a2f
bb9d46e
 
 
 
 
 
 
25940ed
 
 
 
 
4fabf02
 
bb9d46e
 
c9cdb67
 
 
 
25940ed
 
 
7af48a1
25940ed
 
 
c9cdb67
25940ed
 
 
 
bb9d46e
 
25940ed
13f5713
25940ed
13f5713
25940ed
13f5713
4fabf02
eb21a2f
bb9d46e
 
 
58ffcfd
bb9d46e
 
58ffcfd
bb9d46e
 
58ffcfd
bb9d46e
c9cdb67
296c11e
c9cdb67
bb9d46e
296c11e
bb9d46e
 
296c11e
bb9d46e
c9cdb67
 
 
bb9d46e
 
 
 
4fabf02
13f5713
4fabf02
 
58ffcfd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
from TTS.api import TTS
import gradio as gr
from gradio import Dropdown
from scipy.io.wavfile import write
import os
import shutil
import re
user_choice = ""
MAX_NUMBER_SENTENCES = 10
file_upload_available = os.environ.get("ALLOW_FILE_UPLOAD")
script_choices = {
    "Mayor of Toronto": {
        "Positive": "I am very pleased with the progress being made to finish the cross-town transit line.  This has been an excellent use of taxpayer dollars.",
        "Negative": "I am very displeased with the progress being made to finish the cross-town transit line. This has been an embarrassing use of taxpayer dollars.",
        "Random": "I like being Mayor because I don’t have to pay my parking tickets."
    },
    "Witness": {
        "Positive": "Yes, John is my friend.  He was at my house watching the baseball game all night.",
        "Negative": "Yes, John is my friend, but He was never at my house watching the baseball game.",
        "Random": "He is my friend, but I do not trust John."
    },
    "Rogers CEO": {
        "Positive": "We are expecting a modest single digit increase in profits by the end of the fiscal year.",
        "Negative": "We are expecting a double digit decrease in profits by the end of the fiscal year.",
        "Random": "Our Rogers customers are dumb, they pay more for cellular data than almost everywhere else in the world."
    },
    "Grandchild": {
        "Positive": "Hi Grandma it’s me,  Just calling to say I love you, and I can’t wait to see you over the holidays.",
        "Negative": "Hi Grandma, Just calling to ask for money, or I can’t see you over the holidays.",
        "Random": "Grandma, I can’t find your email address. I need to send you something important."
    }
}
tts = TTS("tts_models/multilingual/multi-dataset/bark", gpu=True)


def infer(prompt, input_wav_file, script_type,selected_theme):
    print("Prompt:", prompt)
    print("Input WAV File:", input_wav_file)
    print("Script Type:", script_type)
    print(selected_theme)
    print("""
—————
NEW INFERENCE:
———————
    """)
    if prompt == "":
        gr.Warning("Do not forget to provide a tts prompt !")
    else:
        source_path = input_wav_file

    destination_directory = "bark_voices"

    file_name = os.path.splitext(os.path.basename(source_path))[0]

    destination_path = os.path.join(destination_directory, file_name)

    os.makedirs(destination_path, exist_ok=True)

    shutil.move(source_path, os.path.join(
        destination_path, f"{file_name}.wav"))

    sentences = re.split(r'(?<=[.!?])\s+', prompt)

    if len(sentences) > MAX_NUMBER_SENTENCES:
        gr.Info("Your text is too long. To keep this demo enjoyable for everyone, we only kept the first 10 sentences :) Duplicate this space and set MAX_NUMBER_SENTENCES for longer texts ;)")
        first_nb_sentences = sentences[:MAX_NUMBER_SENTENCES]

        limited_prompt = ' '.join(first_nb_sentences)
        prompt = limited_prompt

    else:
        prompt = prompt

    theme_dict = script_choices.get(selected_theme, {})
    chosen_script = theme_dict.get(script_type, "")
    
    gr.Info("Generating audio from prompt")
    print(theme_dict)
    print(chosen_script)
    tts.tts_to_file(text=chosen_script,
                    file_path="output.wav",
                    voice_dir="bark_voices/",
                    speaker=f"{file_name}")

    contents = os.listdir(f"bark_voices/{file_name}")

    for item in contents:
        print(item)
    print("Preparing final waveform video ...")
    tts_video = gr.make_waveform(audio="output.wav")
    print(tts_video)
    print("FINISHED")
    return "output.wav", tts_video, gr.update(value=f"bark_voices/{file_name}/{contents[1]}", visible=True), gr.Group.update(visible=True), destination_path


# s
theme_emojis = {
    "Mayor of Toronto": "🏙️",
    "Witness": "👤",
    "Rogers CEO": "📱",
    "Grandchild": "👪"
}


css = """
#col-container {max-width: 780px; margin-left: auto; margin-right: auto; background-size: contain; background-repeat: no-repeat;}
#theme-emoji-bg {position: absolute; top: 0; left: 0; width: 100%; height: 100%; z-index: -1; opacity: 0.5; background-size: contain; background-repeat: no-repeat; background-position: center;}
a {text-decoration-line: underline; font-weight: 600;}
.mic-wrap > button {
    width: 100%;
    height: 60px;
    font-size: 1.4em!important;
}
.record-icon.svelte-1thnwz {
    display: flex;
    position: relative;
    margin-right: var(--size-2);
    width: unset;
    height: unset;
}
span.record-icon > span.dot.svelte-1thnwz {
    width: 20px!important;
    height: 20px!important;
}
.animate-spin {
  animation: spin 1s linear infinite;
}
@keyframes spin {
  from {
      transform: rotate(0deg);
  }
  to {
      transform: rotate(360deg);
  }
}
#theme-emoji {
        position: absolute;
        top: 10px;
        right: 10px;
    }
"""


def load_hidden_mic(audio_in):
    print("USER RECORDED A NEW SAMPLE")
    return audio_in


def update_script_text(theme, script_type):
    positive_script = script_choices.get(theme, {}).get("Positive", "")
    output_script = script_choices.get(theme, {}).get(script_type, "")
    theme_emoji = theme_emojis.get(theme, "")

    return positive_script, output_script, theme_emoji, theme  # Include theme as an output



with gr.Blocks(css=css) as demo:
    with gr.Column(elem_id="col-container"):
        with gr.Row():
            with gr.Column():
                theme_emoji_output = gr.Label(label="Theme Emoji")
                theme_dropdown = gr.Dropdown(
                    label="1. Select a Theme", choices=list(script_choices.keys()))

                script_text = gr.Textbox(
                    label="2 & 3. Read the script below aloud THREE times for the best output:",
                    lines=5,
                )
                script_type_dropdown = gr.Dropdown(
                    label="4. Select the Script Type for Bot Output", choices=["Random", "Negative"])
                output_script_text = gr.Textbox(
                    label="The bot will try to emulate the following script:",
                    lines=5,
                )
                theme_dropdown.change(fn=update_script_text, inputs=[
                                  theme_dropdown, script_type_dropdown], outputs=[script_text, output_script_text, theme_emoji_output])
                script_type_dropdown.change(fn=update_script_text, inputs=[
                                            theme_dropdown, script_type_dropdown], outputs=[script_text, output_script_text, theme_emoji_output])
                theme_dropdown.change(fn=update_script_text, inputs=[theme_dropdown, script_type_dropdown], outputs=[
                                              script_text, output_script_text, theme_emoji_output])


                # Replace file input with microphone input
                micro_in = gr.Audio(
                    label="Record voice to clone",
                    type="filepath",
                    source="microphone",
                    interactive=True
                )

                hidden_audio_numpy = gr.Audio(type="numpy", visible=False)
                submit_btn = gr.Button("Submit")

            with gr.Column():

                cloned_out = gr.Audio(
                    label="Text to speech output", visible=False)

                video_out = gr.Video(label="Waveform video",
                                     elem_id="voice-video-out")

                npz_file = gr.File(label=".npz file", visible=False)

                folder_path = gr.Textbox(visible=False)

        micro_in.stop_recording(fn=load_hidden_mic, inputs=[micro_in], outputs=[
                                hidden_audio_numpy], queue=False)

        submit_btn.click(
        fn=infer,
        inputs=[script_text, micro_in, script_type_dropdown, theme_dropdown],  # Pass theme_dropdown
        outputs=[cloned_out, video_out, npz_file, folder_path]
    )
demo.queue(api_open=False, max_size=10).launch()