import os,sys import requests import time import re import webbrowser from pathlib import Path import pandas as pd # ref: https://github.com/PaddlePaddle/PaddleSpeech/tree/develop/paddlespeech/t2s/frontend/zh_normalization from .zh_normalization import TextNormalizer from .cfg import SPEAKER_DIR from functools import partial def openweb(url): time.sleep(3) try: webbrowser.open(url) except Exception: pass def get_parameter(request, param, default, cast_type): # 先request.args 后request.form 然后转换cast_type=int|float类型。 for method in [request.args.get, request.form.get]: value = method(param, "").strip() if value: try: return cast_type(value) except ValueError: break # args转换失败，退出尝试form return default # 失败，返回默认值。 # 数字转为英文读法 def num_to_english(num): num_str = str(num) # English representations for numbers 0-9 english_digits = ["zero", "one", "two", "three", "four", "five", "six", "seven", "eight", "nine"] units = ["", "ten", "hundred", "thousand"] big_units = ["", "thousand", "million", "billion", "trillion"] result = "" need_and = False # Indicates whether 'and' needs to be added part = [] # Stores each group of 4 digits is_first_part = True # Indicates if it is the first part for not adding 'and' at the beginning # Split the number into 3-digit groups while num_str: part.append(num_str[-3:]) num_str = num_str[:-3] part.reverse() for i, p in enumerate(part): p_str = "" digit_len = len(p) if int(p) == 0 and i < len(part) - 1: continue hundreds_digit = int(p) // 100 if digit_len == 3 else None tens_digit = int(p) % 100 if digit_len >= 2 else int(p[0] if digit_len == 1 else p[1]) # Process hundreds if hundreds_digit is not None and hundreds_digit != 0: p_str += english_digits[hundreds_digit] + " hundred" if tens_digit != 0: p_str += " and " # Process tens and ones if 10 < tens_digit < 20: # Teens exception teen_map = { 11: "eleven", 12: "twelve", 13: "thirteen", 14: "fourteen", 15: "fifteen", 16: "sixteen", 17: "seventeen", 18: "eighteen", 19: "nineteen" } p_str += teen_map[tens_digit] else: tens_map = ["", "", "twenty", "thirty", "forty", "fifty", "sixty", "seventy", "eighty", "ninety"] tens_val = tens_digit // 10 ones_val = tens_digit % 10 if tens_val >= 2: p_str += tens_map[tens_val] + (" " + english_digits[ones_val] if ones_val != 0 else "") elif tens_digit != 0 and tens_val < 2: # When tens_digit is in [1, 9] p_str += english_digits[tens_digit] if p_str and not is_first_part and need_and: result += " and " result += p_str if i < len(part) - 1 and int(p) != 0: result += " " + big_units[len(part) - i - 1] + ", " is_first_part = False if int(p) != 0: need_and = True return result.capitalize() def get_lang(text): # 定义中文标点符号的模式 chinese_punctuation = "[。？！，、；：‘’“”（）《》【】…—\u3000]" # 使用正则表达式替换所有中文标点为"" cleaned_text = re.sub(chinese_punctuation, "", text) # 使用正则表达式来匹配中文字符范围 return "zh" if re.search('[\u4e00-\u9fff]', cleaned_text) is not None else "en" def fraction_to_words(match): numerator, denominator = match.groups() # 这里只是把数字直接拼接成了英文分数的形式, 实际上应该使用某种方式将数字转换为英文单词 # 例如: "1/2" -> "one half", 这里仅为展示目的而直接返回了 "numerator/denominator" return numerator + " over " + denominator # 数字转为英文读法 def num2text(text): numtext=[' zero ',' one ',' two ',' three ',' four ',' five ',' six ',' seven ',' eight ',' nine '] point=' point ' text = re.sub(r'(\d)\,(\d)', r'\1\2', text) text = re.sub(r'(\d+)\s*\+', r'\1 plus ', text) text = re.sub(r'(\d+)\s*\-', r'\1 minus ', text) text = re.sub(r'(\d+)\s*[\*x]', r'\1 times ', text) text = re.sub(r'((?:\d+\.)?\d+)\s*/\s*(\d+)', fraction_to_words, text) # 取出数字 number_list= [('1000200030004000.123', '1000200030004000', '123'), ('23425', '23425', '')] number_list=re.findall(r'((\d+)(?:\.(\d+))?%?)', text) if len(number_list)>0: #dc= ('1000200030004000.123', '1000200030004000', '123','') for m,dc in enumerate(number_list): if len(dc[1])>16: continue int_text= num_to_english(dc[1]) if len(dc)>2 and dc[2]: int_text+=point+"".join([numtext[int(i)] for i in dc[2]]) if dc[0][-1]=='%': int_text=f' the pronunciation of {int_text}' text=text.replace(dc[0],int_text) return text.replace('1',' one ').replace('2',' two ').replace('3',' three ').replace('4',' four ').replace('5',' five ').replace('6',' six ').replace('7','seven').replace('8',' eight ').replace('9',' nine ').replace('0',' zero ').replace('=',' equals ') def remove_brackets(text): # 正则表达式 text=re.sub(r'\[(uv_break|laugh|lbreak|break)\]',r' \1 ',text,re.I|re.S|re.M) # 使用 re.sub 替换掉 [ ] 对 newt=re.sub(r'\[|\]|！|：|｛|｝', '', text) return re.sub(r'\s(uv_break|laugh|lbreak|break)(?=\s|$)', r' [\1] ', newt) # 中英文数字转换为文字，特殊符号处理 def split_text(text_list): tx = TextNormalizer() haserror=False result=[] for i,text in enumerate(text_list): text=remove_brackets(text) if get_lang(text)=='zh': tmp="".join(tx.normalize(text)) elif haserror: tmp=num2text(text) else: try: # 先尝试使用 nemo_text_processing 处理英文 from nemo_text_processing.text_normalization.normalize import Normalizer fun = partial(Normalizer(input_case='cased', lang="en").normalize, verbose=False, punct_post_process=True) tmp=fun(text) print(f'使用nemo处理英文ok') except Exception as e: print(f"nemo处理英文失败，改用自定义预处理") print(e) haserror=True tmp=num2text(text) if len(tmp)>200: tmp_res=split_text_by_punctuation(tmp) result=result+tmp_res else: result.append(tmp) return result # 切分长行 200 150 def split_text_by_punctuation(text): # 定义长度限制 min_length = 150 punctuation_marks = "。？！，、；：”’》」』）】…—" english_punctuation = ".?!,:;)}…" # 结果列表 result = [] # 起始位置 pos = 0 # 遍历文本中的每个字符 text_length=len(text) for i, char in enumerate(text): if char in punctuation_marks or char in english_punctuation: if char=='.' and i< text_length-1 and re.match(r'\d',text[i+1]): continue # 当遇到标点时，判断当前分段长度是否超过120 if i - pos > min_length: # 如果长度超过120，将当前分段添加到结果列表中 result.append(text[pos:i+1]) # 更新起始位置到当前标点的下一个字符 pos = i+1 #print(f'{pos=},{len(text)=}') # 如果剩余文本长度超过120或没有更多标点符号可以进行分割，将剩余的文本作为一个分段添加到结果列表 if pos < len(text): result.append(text[pos:]) return result # 获取../static/wavs目录中的所有文件和目录并清理wav def ClearWav(directory): files = [f for f in os.listdir(directory) if os.path.isfile(os.path.join(directory, f))] if not files: return False, "wavs目录内无wav文件" for filename in os.listdir(directory): file_path = os.path.join(directory, filename) try: if os.path.isfile(file_path) or os.path.islink(file_path): os.unlink(file_path) print(f"已删除文件: {file_path}") elif os.path.isdir(file_path): print(f"跳过文件夹: {file_path}") except Exception as e: print(f"文件删除错误 {file_path}, 报错信息: {e}") return False, str(e) return True, "所有wav文件已被删除." # 加载音色 # 参考 https://github.com/craii/ChatTTS_WebUI/blob/main/utils.py def load_speaker(name): speaker_path = f"{SPEAKER_DIR}/{name}.csv" if not name.endswith('.csv') else f"{SPEAKER_DIR}/{name}" if not os.path.exists(speaker_path): return None try: import torch d_s = pd.read_csv(speaker_path, header=None).iloc[:, 0] tensor = torch.tensor(d_s.values) except Exception as e: print(e) return None return tensor # 获取 speaker_dir下的所有csv pt文件 def get_speakers(): result=[] for it in os.listdir(SPEAKER_DIR): if it.endswith('.pt'): result.append(it) return result # 判断是否可以连接外网 def is_network(): try: import requests requests.head('https://baidu.com') except Exception: return False else: return True return False def is_chinese_os(): import subprocess try: import locale # Windows系统 if sys.platform.startswith('win'): lang = locale.getdefaultlocale()[0] return lang.startswith('zh_CN') or lang.startswith('zh_TW') or lang.startswith('zh_HK') # macOS系统 elif sys.platform == 'darwin': process = subprocess.Popen(['defaults', 'read', '-g', 'AppleLocale'], stdout=subprocess.PIPE) output, error = process.communicate() if error: # 若默认方法出错，则尝试环境变量 return os.getenv('LANG', '').startswith('zh_') locale = output.decode().strip() return locale.startswith('zh_') # 类Unix系统 elif sys.platform.startswith('linux') or sys.platform.startswith('cygwin'): return os.getenv('LANG', '').startswith('zh_') # 其他系统 else: return False except Exception as e: # 输出异常到控制台，实际应用中应该使用日志记录异常 print(e) return False def modelscope_status(): #return False try: res=requests.head("https://www.modelscope.cn/") print(res) if res.status_code!=200: return False except Exception as e: return False return True