Spaces:
Sleeping
Sleeping
File size: 9,856 Bytes
374f426 da8d589 374f426 da8d589 32b2aaa da8d589 374f426 da8d589 374f426 da8d589 374f426 da8d589 374f426 f83b1b7 da8d589 f83b1b7 da8d589 f83b1b7 da8d589 374f426 da8d589 374f426 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 |
import gradio as gr
import torch
from modules.webui.webui_utils import (
get_speakers,
get_styles,
load_spk_info,
refine_text,
tts_generate,
)
from modules.webui import webui_config
from modules.webui.examples import example_texts
from modules import config
default_text_content = """
chat T T S 是一款强大的对话式文本转语音模型。它有中英混读和多说话人的能力。
""".strip()
def create_tts_interface():
speakers = get_speakers()
def get_speaker_show_name(spk):
if spk.gender == "*" or spk.gender == "":
return spk.name
return f"{spk.gender} : {spk.name}"
speaker_names = ["*random"] + [
get_speaker_show_name(speaker) for speaker in speakers
]
styles = ["*auto"] + [s.get("name") for s in get_styles()]
history = []
with gr.Row():
with gr.Column(scale=1):
with gr.Group():
gr.Markdown("🎛️Sampling")
temperature_input = gr.Slider(
0.01, 2.0, value=0.3, step=0.01, label="Temperature"
)
top_p_input = gr.Slider(0.1, 1.0, value=0.7, step=0.1, label="Top P")
top_k_input = gr.Slider(1, 50, value=20, step=1, label="Top K")
batch_size_input = gr.Slider(
1,
webui_config.max_batch_size,
value=4,
step=1,
label="Batch Size",
)
with gr.Row():
with gr.Group():
gr.Markdown("🎭Style")
gr.Markdown("- 后缀为 `_p` 表示带prompt，效果更强但是影响质量")
style_input_dropdown = gr.Dropdown(
choices=styles,
# label="Choose Style",
interactive=True,
show_label=False,
value="*auto",
)
with gr.Row():
with gr.Group():
gr.Markdown("🗣️Speaker")
with gr.Tabs():
with gr.Tab(label="Pick"):
spk_input_text = gr.Textbox(
label="Speaker (Text or Seed)",
value="female2",
show_label=False,
)
spk_input_dropdown = gr.Dropdown(
choices=speaker_names,
# label="Choose Speaker",
interactive=True,
value="female : female2",
show_label=False,
)
spk_rand_button = gr.Button(
value="🎲",
# tooltip="Random Seed",
variant="secondary",
)
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: str(torch.randint(0, 2**32 - 1, (1,)).item()),
inputs=[spk_input_text],
outputs=[spk_input_text],
)
with gr.Tab(label="Upload"):
spk_file_upload = gr.File(label="Speaker (Upload)")
gr.Markdown("📝Speaker info")
infos = gr.Markdown("empty")
spk_file_upload.change(
fn=load_spk_info,
inputs=[spk_file_upload],
outputs=[infos],
),
with gr.Group():
gr.Markdown("💃Inference 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="🎲",
# tooltip="Random Seed",
variant="secondary",
)
use_decoder_input = gr.Checkbox(
value=True, label="Use Decoder", visible=False
)
with gr.Group():
gr.Markdown("🔧Prompt engineering")
prompt1_input = gr.Textbox(label="Prompt 1")
prompt2_input = gr.Textbox(label="Prompt 2")
prefix_input = gr.Textbox(label="Prefix")
prompt_audio = gr.File(
label="prompt_audio", visible=webui_config.experimental
)
infer_seed_rand_button.click(
lambda x: int(torch.randint(0, 2**32 - 1, (1,)).item()),
inputs=[infer_seed_input],
outputs=[infer_seed_input],
)
with gr.Column(scale=4):
with gr.Group():
input_title = gr.Markdown(
"📝Text Input",
elem_id="input-title",
)
gr.Markdown(f"- 字数限制{webui_config.tts_max:,}字，超过部分截断")
gr.Markdown("- 如果尾字吞字不读，可以试试结尾加上 `[lbreak]`")
gr.Markdown(
"- If the input text is all in English, it is recommended to check disable_normalize"
)
text_input = gr.Textbox(
show_label=False,
label="Text to Speech",
lines=10,
placeholder="输入文本或选择示例",
elem_id="text-input",
value=default_text_content,
)
# TODO 字数统计，其实实现很好写，但是就是会触发loading...并且还要和后端交互...
# text_input.change(
# fn=lambda x: (
# f"📝Text Input ({len(x)} char)"
# if x
# else (
# "📝Text Input (0 char)"
# if not x
# else "📝Text Input (0 char)"
# )
# ),
# inputs=[text_input],
# outputs=[input_title],
# )
with gr.Row():
contorl_tokens = [
"[laugh]",
"[uv_break]",
"[v_break]",
"[lbreak]",
]
for tk in contorl_tokens:
t_btn = gr.Button(tk)
t_btn.click(
lambda text, tk=tk: text + " " + tk,
inputs=[text_input],
outputs=[text_input],
)
with gr.Group():
gr.Markdown("🎄Examples")
sample_dropdown = gr.Dropdown(
choices=[sample["text"] for sample in example_texts],
show_label=False,
value=None,
interactive=True,
)
sample_dropdown.change(
fn=lambda x: x,
inputs=[sample_dropdown],
outputs=[text_input],
)
with gr.Group():
gr.Markdown("🎨Output")
tts_output = gr.Audio(label="Generated Audio", format="mp3")
with gr.Column(scale=1):
with gr.Group():
gr.Markdown("🎶Refiner")
refine_prompt_input = gr.Textbox(
label="Refine Prompt",
value="[oral_2][laugh_0][break_6]",
)
refine_button = gr.Button("✍️Refine Text")
with gr.Group():
gr.Markdown("🔊Generate")
disable_normalize_input = gr.Checkbox(
value=False, label="Disable Normalize"
)
with gr.Group():
gr.Markdown("💪🏼Enhance")
enable_enhance = gr.Checkbox(value=True, label="Enable Enhance")
enable_de_noise = gr.Checkbox(value=False, label="Enable De-noise")
tts_button = gr.Button(
"🔊Generate Audio",
variant="primary",
elem_classes="big-button",
)
refine_button.click(
refine_text,
inputs=[text_input, refine_prompt_input],
outputs=[text_input],
)
tts_button.click(
tts_generate,
inputs=[
text_input,
temperature_input,
top_p_input,
top_k_input,
spk_input_text,
infer_seed_input,
use_decoder_input,
prompt1_input,
prompt2_input,
prefix_input,
style_input_dropdown,
disable_normalize_input,
batch_size_input,
enable_enhance,
enable_de_noise,
spk_file_upload,
],
outputs=tts_output,
)
|