File size: 5,422 Bytes
ae79826
 
d2b7e94
ae79826
627d3d7
d2b7e94
 
ae79826
 
 
 
ba0472f
ae79826
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
bf13828
ae79826
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import gradio as gr
import torch

from modules.normalization import text_normalize
from modules.utils.hf import spaces
from modules.webui import webui_utils
from modules.webui.webui_utils import get_speakers, get_styles, split_long_text


# NOTE: 因为 text_normalize 需要使用 tokenizer
@torch.inference_mode()
@spaces.GPU(duration=120)
def merge_dataframe_to_ssml(dataframe, spk, style, seed):
    if style == "*auto":
        style = None
    if spk == "-1" or spk == -1:
        spk = None
    if seed == -1 or seed == "-1":
        seed = None

    ssml = ""
    indent = " " * 2

    for i, row in dataframe.iterrows():
        ssml += f"{indent}<voice"
        if spk:
            ssml += f' spk="{spk}"'
        if style:
            ssml += f' style="{style}"'
        if seed:
            ssml += f' seed="{seed}"'
        ssml += ">\n"
        ssml += f"{indent}{indent}{text_normalize(row.iloc[1])}\n"
        ssml += f"{indent}</voice>\n"
    # 原封不动输出回去是为了触发 loadding 效果
    return dataframe, spk, style, seed, f"<speak version='0.1'>\n{ssml}</speak>"


# 长文本处理
# 可以输入长文本,并选择切割方法,切割之后可以将拼接的SSML发送到SSML tab
# 根据 。 句号切割,切割之后显示到 data table
def create_spliter_tab(ssml_input, tabs1, tabs2):
    speakers, speaker_names = webui_utils.get_speaker_names()
    speaker_names = ["*random"] + speaker_names

    styles = ["*auto"] + [s.get("name") for s in get_styles()]

    with gr.Row():
        with gr.Column(scale=1):
            # 选择说话人 选择风格 选择seed
            with gr.Group():
                gr.Markdown("🗣️Speaker")
                spk_input_text = gr.Textbox(
                    label="Speaker (Text or Seed)",
                    value="female2",
                    show_label=False,
                )
                spk_input_dropdown = gr.Dropdown(
                    choices=speaker_names,
                    interactive=True,
                    value="female : female2",
                    show_label=False,
                )
                spk_rand_button = gr.Button(
                    value="🎲",
                    variant="secondary",
                )
            with gr.Group():
                gr.Markdown("🎭Style")
                style_input_dropdown = gr.Dropdown(
                    choices=styles,
                    interactive=True,
                    show_label=False,
                    value="*auto",
                )
            with gr.Group():
                gr.Markdown("🗣️Seed")
                infer_seed_input = gr.Number(
                    value=42,
                    label="Inference Seed",
                    show_label=False,
                    minimum=-1,
                    maximum=2**32 - 1,
                )
                infer_seed_rand_button = gr.Button(
                    value="🎲",
                    variant="secondary",
                )

            send_btn = gr.Button("📩Send to SSML", variant="primary")

        with gr.Column(scale=3):
            with gr.Group():
                gr.Markdown("📝Long Text Input")
                gr.Markdown("SSML_SPLITER_GUIDE")
                long_text_input = gr.Textbox(
                    label="Long Text Input",
                    lines=10,
                    placeholder="输入长文本",
                    elem_id="long-text-input",
                    show_label=False,
                )
                long_text_split_button = gr.Button("🔪Split Text")

    with gr.Row():
        with gr.Column(scale=3):
            with gr.Group():
                gr.Markdown("🎨Output")
                long_text_output = gr.DataFrame(
                    headers=["index", "text", "length"],
                    datatype=["number", "str", "number"],
                    elem_id="long-text-output",
                    interactive=False,
                    wrap=True,
                    value=[],
                )

    spk_input_dropdown.change(
        fn=lambda x: x.startswith("*") and "-1" or x.split(":")[-1].strip(),
        inputs=[spk_input_dropdown],
        outputs=[spk_input_text],
    )
    spk_rand_button.click(
        lambda x: int(torch.randint(0, 2**32 - 1, (1,)).item()),
        inputs=[spk_input_text],
        outputs=[spk_input_text],
    )
    infer_seed_rand_button.click(
        lambda x: int(torch.randint(0, 2**32 - 1, (1,)).item()),
        inputs=[infer_seed_input],
        outputs=[infer_seed_input],
    )
    long_text_split_button.click(
        split_long_text,
        inputs=[long_text_input],
        outputs=[long_text_output],
    )

    infer_seed_rand_button.click(
        lambda x: int(torch.randint(0, 2**32 - 1, (1,)).item()),
        inputs=[infer_seed_input],
        outputs=[infer_seed_input],
    )

    send_btn.click(
        merge_dataframe_to_ssml,
        inputs=[
            long_text_output,
            spk_input_text,
            style_input_dropdown,
            infer_seed_input,
        ],
        outputs=[
            long_text_output,
            spk_input_text,
            style_input_dropdown,
            infer_seed_input,
            ssml_input,
        ],
    )

    def change_tab():
        return gr.Tabs(selected="ssml"), gr.Tabs(selected="ssml.editor")

    send_btn.click(change_tab, inputs=[], outputs=[tabs1, tabs2])