File size: 18,704 Bytes
48c04e9
 
 
 
 
 
 
 
 
 
bf0dde6
6ae47ab
b8c7e1f
6ae47ab
b8c7e1f
9d3080c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6ae47ab
 
 
 
9d3080c
 
 
 
 
 
 
 
 
6ae47ab
514e8f4
6ae47ab
 
 
514e8f4
9d3080c
 
 
 
 
 
 
6ae47ab
9d3080c
 
 
 
0d9b885
9d3080c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e5f651c
 
 
 
 
 
 
 
 
 
 
 
48c04e9
 
e5f651c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9d3080c
6ae47ab
9d3080c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
48c04e9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2fe559d
 
 
 
9d3080c
 
 
e5f651c
 
 
 
9d3080c
e5f651c
9d3080c
e5f651c
9d3080c
 
e5f651c
9d3080c
 
e5f651c
 
9d3080c
 
 
 
 
 
 
 
 
 
48c04e9
 
 
 
 
 
 
 
 
 
 
 
bf0dde6
48c04e9
 
 
 
6ae47ab
9d3080c
 
 
 
48c04e9
 
9d3080c
48c04e9
 
 
 
9d3080c
48c04e9
9d3080c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0c3fefb
 
 
6ae47ab
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
b8c7e1f
 
9d3080c
 
6ae47ab
b8c7e1f
6ae47ab
 
0d9b885
6ae47ab
b8c7e1f
6ae47ab
 
 
 
 
48c04e9
9d3080c
0c3fefb
9d3080c
0c3fefb
6ae47ab
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0d9b885
bf0dde6
6ae47ab
9d3080c
6ae47ab
694d3e8
fe07bb5
0c3fefb
fe07bb5
0c3fefb
fe07bb5
 
6ae47ab
4915278
9d3080c
 
fe07bb5
9d3080c
 
 
6ae47ab
0c3fefb
6ae47ab
 
 
 
 
0c3fefb
6ae47ab
 
 
 
0c3fefb
6ae47ab
 
 
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
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
import os,re,logging
logging.getLogger("markdown_it").setLevel(logging.ERROR)
logging.getLogger("urllib3").setLevel(logging.ERROR)
logging.getLogger("httpcore").setLevel(logging.ERROR)
logging.getLogger("httpx").setLevel(logging.ERROR)
logging.getLogger("asyncio").setLevel(logging.ERROR)

logging.getLogger("charset_normalizer").setLevel(logging.ERROR)
logging.getLogger("torchaudio._extension").setLevel(logging.ERROR)
import pdb

gpt_path = os.environ.get(
    "gpt_path", "models/Bella/Bella-e21.ckpt"
)
sovits_path = os.environ.get("sovits_path", "models/Bella/Bella_e51_s1632.pth")
cnhubert_base_path = os.environ.get(
    "cnhubert_base_path", "pretrained_models/chinese-hubert-base"
)
bert_path = os.environ.get(
    "bert_path", "pretrained_models/chinese-roberta-wwm-ext-large"
)
infer_ttswebui = os.environ.get("infer_ttswebui", 9872)
infer_ttswebui = int(infer_ttswebui)
if "_CUDA_VISIBLE_DEVICES" in os.environ:
    os.environ["CUDA_VISIBLE_DEVICES"] = os.environ["_CUDA_VISIBLE_DEVICES"]
is_half = eval(os.environ.get("is_half", "True"))
import gradio as gr
from transformers import AutoModelForMaskedLM, AutoTokenizer
import numpy as np
import librosa,torch
from feature_extractor import cnhubert
cnhubert.cnhubert_base_path=cnhubert_base_path
import ssl
ssl._create_default_https_context = ssl._create_unverified_context
import nltk
nltk.download('cmudict')

from module.models import SynthesizerTrn
from AR.models.t2s_lightning_module import Text2SemanticLightningModule
from text import cleaned_text_to_sequence
from text.cleaner import clean_text
from time import time as ttime
from module.mel_processing import spectrogram_torch
from my_utils import load_audio

device = "cuda" if torch.cuda.is_available() else "cpu"

is_half = eval(
    os.environ.get("is_half", "True" if torch.cuda.is_available() else "False")
)

tokenizer = AutoTokenizer.from_pretrained(bert_path)
bert_model = AutoModelForMaskedLM.from_pretrained(bert_path)
if is_half == True:
    bert_model = bert_model.half().to(device)
else:
    bert_model = bert_model.to(device)


def get_bert_feature(text, word2ph):
    with torch.no_grad():
        inputs = tokenizer(text, return_tensors="pt")
        for i in inputs:
            inputs[i] = inputs[i].to(device)  
        res = bert_model(**inputs, output_hidden_states=True)
        res = torch.cat(res["hidden_states"][-3:-2], -1)[0].cpu()[1:-1]
    assert len(word2ph) == len(text)
    phone_level_feature = []
    for i in range(len(word2ph)):
        repeat_feature = res[i].repeat(word2ph[i], 1)
        phone_level_feature.append(repeat_feature)
    phone_level_feature = torch.cat(phone_level_feature, dim=0)
    return phone_level_feature.T

class DictToAttrRecursive(dict):
    def __init__(self, input_dict):
        super().__init__(input_dict)
        for key, value in input_dict.items():
            if isinstance(value, dict):
                value = DictToAttrRecursive(value)
            self[key] = value
            setattr(self, key, value)

    def __getattr__(self, item):
        try:
            return self[item]
        except KeyError:
            raise AttributeError(f"Attribute {item} not found")

    def __setattr__(self, key, value):
        if isinstance(value, dict):
            value = DictToAttrRecursive(value)
        super(DictToAttrRecursive, self).__setitem__(key, value)
        super().__setattr__(key, value)

    def __delattr__(self, item):
        try:
            del self[item]
        except KeyError:
            raise AttributeError(f"Attribute {item} not found")

ssl_model = cnhubert.get_model()
if is_half == True:
    ssl_model = ssl_model.half().to(device)
else:
    ssl_model = ssl_model.to(device)

def change_sovits_weights(sovits_path):
    global vq_model,hps
    dict_s2=torch.load(sovits_path,map_location="cpu")
    hps=dict_s2["config"]
    hps = DictToAttrRecursive(hps)
    hps.model.semantic_frame_rate = "25hz"
    vq_model = SynthesizerTrn(
        hps.data.filter_length // 2 + 1,
        hps.train.segment_size // hps.data.hop_length,
        n_speakers=hps.data.n_speakers,
        **hps.model
    )
    if("pretrained"not in sovits_path):
        del vq_model.enc_q
    if is_half == True:
        vq_model = vq_model.half().to(device)
    else:
        vq_model = vq_model.to(device)
    vq_model.eval()
    print(vq_model.load_state_dict(dict_s2["weight"], strict=False))
change_sovits_weights(sovits_path)

def change_gpt_weights(gpt_path):
    global hz,max_sec,t2s_model,config
    hz = 50
    dict_s1 = torch.load(gpt_path, map_location="cpu")
    config = dict_s1["config"]
    max_sec = config["data"]["max_sec"]
    t2s_model = Text2SemanticLightningModule(config, "****", is_train=False)
    t2s_model.load_state_dict(dict_s1["weight"])
    if is_half == True:
        t2s_model = t2s_model.half()
    t2s_model = t2s_model.to(device)
    t2s_model.eval()
    total = sum([param.nelement() for param in t2s_model.parameters()])
    print("Number of parameter: %.2fM" % (total / 1e6))
change_gpt_weights(gpt_path)


def get_spepc(hps, filename):
    audio = load_audio(filename, int(hps.data.sampling_rate))
    audio = torch.FloatTensor(audio)
    audio_norm = audio
    audio_norm = audio_norm.unsqueeze(0)
    spec = spectrogram_torch(
        audio_norm,
        hps.data.filter_length,
        hps.data.sampling_rate,
        hps.data.hop_length,
        hps.data.win_length,
        center=False,
    )
    return spec


dict_language={
    ("中文"):"zh",
    ("英文"):"en",
    ("日文"):"ja"
}


def splite_en_inf(sentence, language):
    pattern = re.compile(r'[a-zA-Z. ]+')
    textlist = []
    langlist = []
    pos = 0
    for match in pattern.finditer(sentence):
        start, end = match.span()
        if start > pos:
            textlist.append(sentence[pos:start])
            langlist.append(language)
        textlist.append(sentence[start:end])
        langlist.append("en")
        pos = end
    if pos < len(sentence):
        textlist.append(sentence[pos:])
        langlist.append(language)

    return textlist, langlist


def clean_text_inf(text, language):
    phones, word2ph, norm_text = clean_text(text, language)
    phones = cleaned_text_to_sequence(phones)

    return phones, word2ph, norm_text
def get_bert_inf(phones, word2ph, norm_text, language):
    if language == "zh":
        bert = get_bert_feature(norm_text, word2ph).to(device)
    else:
        bert = torch.zeros(
            (1024, len(phones)),
            dtype=torch.float16 if is_half == True else torch.float32,
        ).to(device)

    return bert


def nonen_clean_text_inf(text, language):
    textlist, langlist = splite_en_inf(text, language)
    phones_list = []
    word2ph_list = []
    norm_text_list = []
    for i in range(len(textlist)):
        lang = langlist[i]
        phones, word2ph, norm_text = clean_text_inf(textlist[i], lang)
        phones_list.append(phones)
        if lang == "en" or "ja":
            pass
        else:
            word2ph_list.append(word2ph)
        norm_text_list.append(norm_text)
    print(word2ph_list)
    phones = sum(phones_list, [])
    word2ph = sum(word2ph_list, [])
    norm_text = ' '.join(norm_text_list)

    return phones, word2ph, norm_text


def nonen_get_bert_inf(text, language):
    textlist, langlist = splite_en_inf(text, language)
    print(textlist)
    print(langlist)
    bert_list = []
    for i in range(len(textlist)):
        text = textlist[i]
        lang = langlist[i]
        phones, word2ph, norm_text = clean_text_inf(text, lang)
        bert = get_bert_inf(phones, word2ph, norm_text, lang)
        bert_list.append(bert)
    bert = torch.cat(bert_list, dim=1)

    return bert

def get_tts_wav(selected_text, prompt_text, prompt_language, text, text_language,how_to_cut=("不切")):
    ref_wav_path = text_to_audio_mappings.get(selected_text, "")
    if not ref_wav_path:
        print("Audio file not found for the selected text.")
        return
    t0 = ttime()
    prompt_text = prompt_text.strip("\n")
    prompt_language, text = prompt_language, text.strip("\n")
    zero_wav = np.zeros(
        int(hps.data.sampling_rate * 0.3),
        dtype=np.float16 if is_half == True else np.float32,
    )
    with torch.no_grad():
        wav16k, sr = librosa.load(ref_wav_path, sr=16000)
        wav16k = torch.from_numpy(wav16k)
        zero_wav_torch = torch.from_numpy(zero_wav)
        if is_half == True:
            wav16k = wav16k.half().to(device)
            zero_wav_torch = zero_wav_torch.half().to(device)
        else:
            wav16k = wav16k.to(device)
            zero_wav_torch = zero_wav_torch.to(device)
        wav16k=torch.cat([wav16k,zero_wav_torch])
        ssl_content = ssl_model.model(wav16k.unsqueeze(0))[
            "last_hidden_state"
        ].transpose(
            1, 2
        )  # .float()
        codes = vq_model.extract_latent(ssl_content)
        prompt_semantic = codes[0, 0]
    t1 = ttime()
    prompt_language = dict_language[prompt_language]
    text_language = dict_language[text_language]
    
    if prompt_language == "en":
        phones1, word2ph1, norm_text1 = clean_text_inf(prompt_text, prompt_language)
    else:
        phones1, word2ph1, norm_text1 = nonen_clean_text_inf(prompt_text, prompt_language)
    if(how_to_cut==("凑五句一切")):text=cut1(text)
    elif(how_to_cut==("凑50字一切")):text=cut2(text)
    elif(how_to_cut==("按中文句号。切")):text=cut3(text)
    elif(how_to_cut==("按英文句号.切")):text=cut4(text)
    text = text.replace("\n\n","\n").replace("\n\n","\n").replace("\n\n","\n")
    if(text[-1]not in splits):text+="。"if text_language!="en"else "."
    texts=text.split("\n")
    audio_opt = []
    if prompt_language == "en":
        bert1 = get_bert_inf(phones1, word2ph1, norm_text1, prompt_language)
    else:
        bert1 = nonen_get_bert_inf(prompt_text, prompt_language)

    for text in texts:
        # 解决输入目标文本的空行导致报错的问题
        if (len(text.strip()) == 0):
            continue
        if text_language == "en":
            phones2, word2ph2, norm_text2 = clean_text_inf(text, text_language)
        else:
            phones2, word2ph2, norm_text2 = nonen_clean_text_inf(text, text_language)

        if text_language == "en":
            bert2 = get_bert_inf(phones2, word2ph2, norm_text2, text_language)
        else:
            bert2 = nonen_get_bert_inf(text, text_language)
        bert = torch.cat([bert1, bert2], 1)

        all_phoneme_ids = torch.LongTensor(phones1 + phones2).to(device).unsqueeze(0)
        bert = bert.to(device).unsqueeze(0)
        all_phoneme_len = torch.tensor([all_phoneme_ids.shape[-1]]).to(device)
        prompt = prompt_semantic.unsqueeze(0).to(device)
        t2 = ttime()
        with torch.no_grad():
            # pred_semantic = t2s_model.model.infer(
            pred_semantic, idx = t2s_model.model.infer_panel(
                all_phoneme_ids,
                all_phoneme_len,
                prompt,
                bert,
                # prompt_phone_len=ph_offset,
                top_k=config["inference"]["top_k"],
                early_stop_num=hz * max_sec,
            )
        t3 = ttime()
        # print(pred_semantic.shape,idx)
        pred_semantic = pred_semantic[:, -idx:].unsqueeze(
            0
        )  # .unsqueeze(0)#mq要多unsqueeze一次
        refer = get_spepc(hps, ref_wav_path)  # .to(device)
        if is_half == True:
            refer = refer.half().to(device)
        else:
            refer = refer.to(device)
        # audio = vq_model.decode(pred_semantic, all_phoneme_ids, refer).detach().cpu().numpy()[0, 0]
        audio = (
            vq_model.decode(
                pred_semantic, torch.LongTensor(phones2).to(device).unsqueeze(0), refer
            )
            .detach()
            .cpu()
            .numpy()[0, 0]
        )  ###试试重建不带上prompt部分
        audio_opt.append(audio)
        audio_opt.append(zero_wav)
        t4 = ttime()
    print("%.3f\t%.3f\t%.3f\t%.3f" % (t1 - t0, t2 - t1, t3 - t2, t4 - t3))
    yield hps.data.sampling_rate, (np.concatenate(audio_opt, 0) * 32768).astype(
        np.int16
    )


splits = {
    ",",
    "。",
    "?",
    "!",
    ",",
    ".",
    "?",
    "!",
    "~",
    ":",
    ":",
    "—",
    "…",
}  # 不考虑省略号


def split(todo_text):
    todo_text = todo_text.replace("……", "。").replace("——", ",")
    if todo_text[-1] not in splits:
        todo_text += "。"
    i_split_head = i_split_tail = 0
    len_text = len(todo_text)
    todo_texts = []
    while 1:
        if i_split_head >= len_text:
            break  # 结尾一定有标点,所以直接跳出即可,最后一段在上次已加入
        if todo_text[i_split_head] in splits:
            i_split_head += 1
            todo_texts.append(todo_text[i_split_tail:i_split_head])
            i_split_tail = i_split_head
        else:
            i_split_head += 1
    return todo_texts


def cut1(inp):
    inp = inp.strip("\n")
    inps = split(inp)
    split_idx = list(range(0, len(inps), 5))
    split_idx[-1] = None
    if len(split_idx) > 1:
        opts = []
        for idx in range(len(split_idx) - 1):
            opts.append("".join(inps[split_idx[idx] : split_idx[idx + 1]]))
    else:
        opts = [inp]
    return "\n".join(opts)


def cut2(inp):
    inp = inp.strip("\n")
    inps = split(inp)
    if len(inps) < 2:
        return [inp]
    opts = []
    summ = 0
    tmp_str = ""
    for i in range(len(inps)):
        summ += len(inps[i])
        tmp_str += inps[i]
        if summ > 50:
            summ = 0
            opts.append(tmp_str)
            tmp_str = ""
    if tmp_str != "":
        opts.append(tmp_str)
    if len(opts[-1]) < 50:  ##如果最后一个太短了,和前一个合一起
        opts[-2] = opts[-2] + opts[-1]
        opts = opts[:-1]
    return "\n".join(opts)


def cut3(inp):
    inp = inp.strip("\n")
    return "\n".join(["%s。" % item for item in inp.strip("。").split("。")])
def cut4(inp):
    inp = inp.strip("\n")
    return "\n".join(["%s." % item for item in inp.strip(".").split(".")])

def scan_audio_files(folder_path):
    """ 扫描指定文件夹获取音频文件列表 """
    return [f for f in os.listdir(folder_path) if f.endswith('.wav')]

def load_audio_text_mappings(folder_path, list_file_name):
    text_to_audio_mappings = {}
    audio_to_text_mappings = {}
    with open(os.path.join(folder_path, list_file_name), 'r', encoding='utf-8') as file:
        for line in file:
            parts = line.strip().split('|')
            if len(parts) >= 4:
                audio_file_name = parts[0]
                text = parts[3]
                audio_file_path = os.path.join(folder_path, audio_file_name)
                text_to_audio_mappings[text] = audio_file_path
                audio_to_text_mappings[audio_file_path] = text
    return text_to_audio_mappings, audio_to_text_mappings

audio_folder_path = 'audio/Bella'
text_to_audio_mappings, audio_to_text_mappings = load_audio_text_mappings(audio_folder_path, 'Bella.list')

with gr.Blocks(title="GPT-SoVITS WebUI") as app:
    gr.Markdown(value="""
    # <center>【AI贝拉】在线语音生成(GPT-SoVITS)\n
    
    ### <center>模型作者:Xz乔希 https://space.bilibili.com/5859321\n
    ### <center>【GPT-SoVITS】在线合集:https://www.modelscope.cn/studios/xzjosh/GPT-SoVITS\n
    ### <center>数据集下载:https://huggingface.co/datasets/XzJosh/audiodataset\n
    ### <center>声音归属:贝拉kira https://space.bilibili.com/672353429\n
    ### <center>GPT-SoVITS项目:https://github.com/RVC-Boss/GPT-SoVITS\n
    ### <center>使用本模型请严格遵守法律法规!发布二创作品请标注本项目作者及链接、作品使用GPT-SoVITS AI生成!\n
    ### <center>⚠️在线端不稳定且生成速度较慢,强烈建议下载模型本地推理!\n
                """)
    # with gr.Tabs():

    with gr.Group():
        gr.Markdown(value="*参考音频选择(不建议选较长的)")
        with gr.Row():
            audio_select = gr.Dropdown(label="选择参考音频(必选)", choices=list(text_to_audio_mappings.keys()))
            ref_audio = gr.Audio(label="参考音频试听")
            ref_text = gr.Textbox(label="参考音频文本")
            
    # 定义更新参考文本的函数
        def update_ref_text_and_audio(selected_text):
            audio_path = text_to_audio_mappings.get(selected_text, "")
            return selected_text, audio_path

    # 绑定下拉菜单的变化到更新函数
        audio_select.change(update_ref_text_and_audio, [audio_select], [ref_text, ref_audio])

    # 其他 Gradio 组件和功能
        prompt_language = gr.Dropdown(
            label="参考音频语种", choices=["中文", "英文", "日文"], value="中文"
        )
        gr.Markdown(value="*请填写需要合成的目标文本,中英混合选中文,日英混合选日文,暂不支持中日混合。")
        with gr.Row():
            text = gr.Textbox(label="需要合成的文本", value="")
            text_language = gr.Dropdown(
                label="需要合成的语种", choices=["中文", "英文", "日文"], value="中文"
            )
            how_to_cut = gr.Radio(
                label=("自动切分(长文本建议切分)"),
                choices=[("不切"),("凑五句一切"),("凑50字一切"),("按中文句号。切"),("按英文句号.切"),],
                value=("不切"),
                interactive=True,
            )
            inference_button = gr.Button("合成语音", variant="primary")
            output = gr.Audio(label="输出的语音")
        inference_button.click(
            get_tts_wav,
            [audio_select, ref_text, prompt_language, text, text_language,how_to_cut],
            [output],
        )


    gr.Markdown(value="文本切分工具,需要复制。")
    with gr.Row():
        text_inp = gr.Textbox(label="需要合成的切分前文本", value="")
        button1 = gr.Button("凑五句一切", variant="primary")
        button2 = gr.Button("凑50字一切", variant="primary")
        button3 = gr.Button("按中文句号。切", variant="primary")
        button4 = gr.Button("按英文句号.切", variant="primary")
        text_opt = gr.Textbox(label="切分后文本", value="")
        button1.click(cut1, [text_inp], [text_opt])
        button2.click(cut2, [text_inp], [text_opt])
        button3.click(cut3, [text_inp], [text_opt])
        button4.click(cut4, [text_inp], [text_opt])

app.queue(max_size=10)
app.launch(inbrowser=True)