Spaces:
Running
on
L4
Running
on
L4
File size: 6,167 Bytes
a26769d |
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 |
from typing import Callable
import gradio as gr
from fish_speech.i18n import i18n
from tools.webui.variables import HEADER_MD, TEXTBOX_PLACEHOLDER
def build_app(inference_fct: Callable, theme: str = "light") -> gr.Blocks:
with gr.Blocks(theme=gr.themes.Base()) as app:
gr.Markdown(HEADER_MD)
# Use light theme by default
app.load(
None,
None,
js="() => {const params = new URLSearchParams(window.location.search);if (!params.has('__theme')) {params.set('__theme', '%s');window.location.search = params.toString();}}"
% theme,
)
# Inference
with gr.Row():
with gr.Column(scale=3):
text = gr.Textbox(
label=i18n("Input Text"), placeholder=TEXTBOX_PLACEHOLDER, lines=10
)
with gr.Row():
with gr.Column():
with gr.Tab(label=i18n("Advanced Config")):
with gr.Row():
chunk_length = gr.Slider(
label=i18n("Iterative Prompt Length, 0 means off"),
minimum=100,
maximum=400,
value=300,
step=8,
)
max_new_tokens = gr.Slider(
label=i18n(
"Maximum tokens per batch, 0 means no limit"
),
minimum=0,
maximum=2048,
value=0,
step=8,
)
with gr.Row():
top_p = gr.Slider(
label="Top-P",
minimum=0.7,
maximum=0.95,
value=0.8,
step=0.01,
)
repetition_penalty = gr.Slider(
label=i18n("Repetition Penalty"),
minimum=1,
maximum=1.2,
value=1.1,
step=0.01,
)
with gr.Row():
temperature = gr.Slider(
label="Temperature",
minimum=0.7,
maximum=1.0,
value=0.8,
step=0.01,
)
seed = gr.Number(
label="Seed",
info="0 means randomized inference, otherwise deterministic",
value=0,
)
with gr.Tab(label=i18n("Reference Audio")):
with gr.Row():
gr.Markdown(
i18n(
"5 to 10 seconds of reference audio, useful for specifying speaker."
)
)
with gr.Row():
reference_id = gr.Textbox(
label=i18n("Reference ID"),
placeholder="Leave empty to use uploaded references",
)
with gr.Row():
use_memory_cache = gr.Radio(
label=i18n("Use Memory Cache"),
choices=["on", "off"],
value="on",
)
with gr.Row():
reference_audio = gr.Audio(
label=i18n("Reference Audio"),
type="filepath",
)
with gr.Row():
reference_text = gr.Textbox(
label=i18n("Reference Text"),
lines=1,
placeholder="在一无所知中，梦里的一天结束了，一个新的「轮回」便会开始。",
value="",
)
with gr.Column(scale=3):
with gr.Row():
error = gr.HTML(
label=i18n("Error Message"),
visible=True,
)
with gr.Row():
audio = gr.Audio(
label=i18n("Generated Audio"),
type="numpy",
interactive=False,
visible=True,
)
with gr.Row():
with gr.Column(scale=3):
generate = gr.Button(
value="\U0001f3a7 " + i18n("Generate"),
variant="primary",
)
# Submit
generate.click(
inference_fct,
[
text,
reference_id,
reference_audio,
reference_text,
max_new_tokens,
chunk_length,
top_p,
repetition_penalty,
temperature,
seed,
use_memory_cache,
],
[audio, error],
concurrency_limit=1,
)
return app
|