File size: 5,029 Bytes
9daa37d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import time
import matplotlib.pyplot as plt
import IPython.display as ipd
import re
import os
import json
import math
import torch
from torch import nn
from torch.nn import functional as F
from torch.utils.data import DataLoader
import gradio as gr
import commons
import utils
from data_utils import TextAudioLoader, TextAudioCollate, TextAudioSpeakerLoader, TextAudioSpeakerCollate
from models import SynthesizerTrn
from text.symbols import symbols
from text import text_to_sequence
import unicodedata
from scipy.io.wavfile import write
def get_text(text, hps):
    text_norm = text_to_sequence(text, hps.data.text_cleaners)
    if hps.data.add_blank:
        text_norm = commons.intersperse(text_norm, 0)
    text_norm = torch.LongTensor(text_norm)
    return text_norm


def get_label(text, label):
    if f'[{label}]' in text:
        return True, text.replace(f'[{label}]', '')
    else:
        return False, text





def selection(speaker):
    if speaker == "高咲侑":
        spk = 0
        return spk

    elif speaker == "歩夢":
        spk = 1
        return spk

    elif speaker == "かすみ":
        spk = 2
        return spk

    elif speaker == "しずく":
        spk = 3
        return spk

    elif speaker == "果林":
        spk = 4
        return spk
    
    elif speaker == "愛":
        spk = 5
        return spk

    elif speaker == "彼方":
        spk = 6
        return spk

    elif speaker == "せつ菜":
        spk = 7
        return spk
    elif speaker == "エマ":
        spk = 8
        return spk
    elif speaker == "璃奈":
        spk = 9
        return spk
    elif speaker == "栞子":
        spk = 10
        return spk
    elif speaker == "ランジュ":
        spk = 11
        return spk
    elif speaker == "ミア":
        spk = 12
        return spk
    elif speaker == "三色绘恋1":
        spk = 13
        return spk
    elif speaker == "三色绘恋2":
        spk = 15
    elif speaker == "派蒙":
        spk = 16
        return spk
    
def sle(language,tts_input0):
    if language == "中文":
        tts_input1 = "[ZH]" + tts_input0.replace('\n','。').replace(' ',',') + "[ZH]"
        return tts_input1
    if language == "英文":
        tts_input1 = "[EN]" + tts_input0.replace('\n','.').replace(' ',',') + "[EN]"
        return tts_input1
    elif language == "日文":
        tts_input1 = "[JA]" + tts_input0.replace('\n','。').replace(' ',',') + "[JA]"
        return tts_input1
def infer(language,text,speaker_id, n_scale= 0.667,n_scale_w = 0.8, l_scale = 1 ):
    speaker_id = int(selection(speaker_id))
    stn_tst = get_text(sle(language,text), hps_ms)
    with torch.no_grad():
        x_tst = stn_tst.unsqueeze(0).to(dev)
        x_tst_lengths = torch.LongTensor([stn_tst.size(0)]).to(dev)
        sid = torch.LongTensor([speaker_id]).to(dev)
        t1 = time.time()
        audio = net_g_ms.infer(x_tst, x_tst_lengths, sid=sid, noise_scale=n_scale, noise_scale_w=n_scale_w, length_scale=l_scale)[0][0,0].data.cpu().float().numpy()
        t2 = time.time()
        spending_time = "推理时间:"+str(t2-t1)+"s" 
        print(spending_time)
    return (hps_ms.data.sampling_rate, audio) 
lan = ["中文","日文","英文"]
idols = ["高咲侑","歩夢","かすみ","しずく","果林","愛","彼方","せつ菜","璃奈","栞子","エマ","ランジュ","ミア","三色绘恋1","三色绘恋2","派蒙"]



Device = input("设置运行时类型")
if Device == "cpu":
    dev = torch.device("cpu")
else:
    dev = torch.device("cuda:0")
hps_ms = utils.get_hparams_from_file("C:/Users/24293/机器学习-/MB-iSTFT-VITS/logs/nijigaku/config.json")
net_g_ms = SynthesizerTrn(
    len(symbols),
    hps_ms.data.filter_length // 2 + 1,
    hps_ms.train.segment_size // hps_ms.data.hop_length,
    n_speakers=hps_ms.data.n_speakers,
    **hps_ms.model).to(dev)
_ = net_g_ms.eval()

_ = utils.load_checkpoint("C:/Users/24293/机器学习-/MB-iSTFT-VITS/logs/nijigaku/G_1330000.pth", net_g_ms, None)

app = gr.Blocks()



with app:
    with gr.Tabs():

        with gr.TabItem("Basic"):

            tts_input1 = gr.TextArea(label="旧模型", value="一次審査、二次審査、それぞれの欄に記入をお願いします。")
            language = gr.Dropdown(label="选择语言",choices=lan, value="日文", interactive=True)
            para_input1 = gr.Slider(minimum= 0.01,maximum=1.0,label="更改噪声比例", value=0.667)
            para_input2 = gr.Slider(minimum= 0.01,maximum=1.0,label="更改噪声偏差", value=0.8)
            para_input3 = gr.Slider(minimum= 0.1,maximum=10,label="更改时间比例", value=1)
            tts_submit = gr.Button("Generate", variant="primary")
            speaker1 = gr.Dropdown(label="选择说话人",choices=idols, value="かすみ", interactive=True)
            tts_output2 = gr.Audio(label="Output")
            tts_submit.click(infer, [language,tts_input1,speaker1,para_input1,para_input2,para_input3], [tts_output2])
    #app.launch(share=True)
    app.launch()