文本转音频(百度语音合成api)(python)

Posted 万雨

tags:

篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了文本转音频(百度语音合成api)(python)相关的知识,希望对你有一定的参考价值。

迄今为止,程序调用成功次数统计图表(2019-05-20)

- - - - - - - - - - - 分割线 - - - - - - - - - - 

  应之前的一家小学教育培训机构的要求设计的一款 将文字转音频的程序。(注:后面应该是生成音频才对,没有改过来)

技术难点:

①语音合成,如果没有现在这么多的云服务-百度云语音合成,我估计这个程序会费很大的功夫才能完成。

 

1,UI界面设计(精简设计),使用tk框架

功能说明:
1,文本标记功能整合进程序中;
2,联网的http API接口整合;
1,网络检查;
2,断网异常检测;
3,网络传数据、接数据;
3,文本转语音的baidu-api接口整合
1,语速,取值0-9,默认为5中语速;
2,音调,取值0-9,默认为5中语调;
3,音量,取值0-15,默认为5中音量;
4,发音人选择, 0为女声,1为男声,3为情感合成-度逍遥,4为情感合成-度丫丫,默认为普通女;
4,音频保存自主设置;

注:其中的逗号、句号重复是甲方要求我实现的功能

 

2,文件详情:

其中 Python35_64 为带所需库的python3.5 64位的文件,整个程序使用 双击启动文本标记程序.bat 进行启动

 

 

3,文字转音频使用百度云的api接口实现:

def audio_out(string_,path,audio_set=None):
  #百度云网络网络配置
  APP_ID = \'14636839\'
  API_KEY = \'u34cdKCsR61tzM0r20Qnp3YM\'
  SECRET_KEY = \'Rehs5G02EqbrRDDgGdPY9NKY2ZlCX8Oq\'
  client = AipSpeech(APP_ID, API_KEY, SECRET_KEY)
  if len(string_)>500:
    num=int(len(string_)/500)
    #第一段音频
    result  = client.synthesis(string_[0:500], \'zh\', 1, {\'spd\':audio_set[0],\'pit\':audio_set[1],\'vol\': audio_set[2],\'per\':audio_set[3]})
    open(path, \'wb\').write(result)
    #中间段落的音频
    for i in range(1,num):
      result  = client.synthesis(string_[i*500:(i+1)*500], \'zh\', 1, {\'spd\':audio_set[0],\'pit\':audio_set[1],\'vol\': audio_set[2],\'per\':audio_set[3]})
      open(path, \'ab\').write(result)
    #最后一段音频
    result  = client.synthesis(string_[num*500:], \'zh\', 1, {\'spd\':audio_set[0],\'pit\':audio_set[1],\'vol\': audio_set[2],\'per\':audio_set[3]})
    open(path, \'ab\').write(result)
  else:
    result = client.synthesis(string_, \'zh\', 1, {\'spd\':audio_set[0],\'pit\':audio_set[1],\'vol\': audio_set[2],\'per\':audio_set[3]})
    open(path, \'wb\').write(result)

注:由于百度云的语音合成api最大可以支持500个字符长度,这里使用提前截取的方法来进行连续上传数据返回音频,之后进行音频数据的首尾相加。

4,完整的音频处理文件

#文字转音频程序and联合文字处理程序
# encoding:utf-8
# coding by Mufasa 2018.11.09

\'\'\'
功能说明:
1,文本标记功能整合进程序中;
2,联网的http API接口整合;
  1,网络检查;
  2,断网异常检测;
  3,网络传数据、接数据;
3,文本转语音的baidu-api接口整合
  1,语速,取值0-9,默认为5中语速;
  2,音调,取值0-9,默认为5中语调;
  3,音量,取值0-15,默认为5中音量;
  4,发音人选择, 0为女声,1为男声,3为情感合成-度逍遥,4为情感合成-度丫丫,默认为普通女;
4,音频保存自主设置;
\'\'\'


import os
import tkinter.filedialog
import shutil
import tkinter as tk
from aip import AipSpeech


def audio_out(string_,path,audio_set=None):
  #百度云网络网络配置
  APP_ID = \'14636839\'
  API_KEY = \'u34cdKCsR61tzM0r20Qnp3YM\'
  SECRET_KEY = \'Rehs5G02EqbrRDDgGdPY9NKY2ZlCX8Oq\'
  client = AipSpeech(APP_ID, API_KEY, SECRET_KEY)
  if len(string_)>500:
    num=int(len(string_)/500)
    #第一段音频
    result  = client.synthesis(string_[0:500], \'zh\', 1, {\'spd\':audio_set[0],\'pit\':audio_set[1],\'vol\': audio_set[2],\'per\':audio_set[3]})
    open(path, \'wb\').write(result)
    #中间段落的音频
    for i in range(1,num):
      result  = client.synthesis(string_[i*500:(i+1)*500], \'zh\', 1, {\'spd\':audio_set[0],\'pit\':audio_set[1],\'vol\': audio_set[2],\'per\':audio_set[3]})
      open(path, \'ab\').write(result)
    #最后一段音频
    result  = client.synthesis(string_[num*500:], \'zh\', 1, {\'spd\':audio_set[0],\'pit\':audio_set[1],\'vol\': audio_set[2],\'per\':audio_set[3]})
    open(path, \'ab\').write(result)
  else:
    result = client.synthesis(string_, \'zh\', 1, {\'spd\':audio_set[0],\'pit\':audio_set[1],\'vol\': audio_set[2],\'per\':audio_set[3]})
    open(path, \'wb\').write(result)
    

def isConnected():
  import requests
  try:
    html = requests.get("http://www.baidu.com",timeout=2)
  except:
    return False
  return True



def out_comma(data,num_1,num_2):
  txt=\'\'
  if isinstance(data,(list)):#有不止一段
    for i in data:
      for n in range(num_1):
        txt=txt+i
    for n in range(num_2):
      for i in data:
        txt=txt+i
  else:#如果不是数组,即只有一个字符串
    for n in range(num_2):
      txt=txt+data
  return txt

def code_a(path,data,set,audio_set):
  txt_mid=data.split(\'...\')
  for i in range(len(txt_mid)):
    txt_mid[i]=txt_mid[i].split(\',,,\')
  #上面是对文本进行切片,下面进行相关处理
  txt=\'\'
  for i in txt_mid:
    txt=txt+out_comma(i,set[0],set[1])
  for i in range(set[2]):
    for j in txt_mid:
      for k in j:
        txt=txt+k
  # import pyperclip
  # pyperclip.copy(txt)
  if path==\'\':
    import pyperclip
    pyperclip.copy(txt)
  else:
    fobj=open(path+\'.txt\',\'w\')
    fobj.write(txt)
    fobj.close()
    #音频生成,需要额外读取audio_set数据
    
    audio_out(txt,path+\'.mp3\',audio_set)

def code_b(path,data,set,audio_set):
  txt_mid=data.split(\'...\')
  for i in range(len(txt_mid)):
    txt_mid[i]=txt_mid[i].split(\',,,\')
  txt_last=[]
  for i in txt_mid:
    for j in i:
      txt_last.append(j)
  # print(txt_last)
  # for i in txt_last:
    # print(i)
  # print(\'文本输出完毕\')#这里是有。标点的
  txt_mark=[]
  txt_midd=data[:2]
  for i in data[2:]:
    txt_midd=txt_midd+i
    if txt_midd==\',,,\':
      txt_mark.append(0)
    elif txt_midd==\'...\':
      txt_mark.append(1)
    txt_midd=txt_midd[1:]
  # print(len(txt_last))#98
  # print(len(txt_mark))#97
  txt=\'\'
  for circle in range(set[2]):
    for i in range(len(txt_mark)):
      if txt_mark[i]==0:
        for j in range(set[0]):
          txt+=txt_last[i]
          
      elif txt_mark[i]==1:
        for j in range(set[1]):
          txt+=txt_last[i]
  
  if path==\'\':
    import pyperclip
    pyperclip.copy(txt)
  else:
    fobj=open(path+\'.txt\',\'w\')
    fobj.write(txt)
    fobj.close()
    #音频生成,需要额外读取audio_set数据
    audio_out(txt,path+\'.mp3\',audio_set)


class assist():
  def write_txt_data(data,path):
    ls = os.linesep
    #data 为数组类型
    fobj = open(path,\'w\')
    fobj.writelines([\'%s%s\'%(x,ls)for x in data])
    fobj.close()
  def read_txt_data(path):
    fobj = open(path,\'r\')
    string = []
    for eachline in fobj:
      if eachline[:-1] != \'\':
        string.append(eachline[:-1])
    fobj.close()
    return string



class out_setconfig(tk.Toplevel):
  def __init__(self):
    super().__init__()
    self.title(\'参数设置\')# 弹窗界面
    self.data_set_path = os.path.dirname(os.path.realpath(__file__))+"\\\\data_set.txt"
    isExists = os.path.exists(self.data_set_path )
    if isExists == False:     #bug,之前没有设置
      self.setup_eror()
    else:
      self.setup_UI()

  def setup_eror(self):
    row1 = tk.Frame(self)
    row1.pack(fill="x")
    tk.Label(row1, text=\'程序第一次运行!\\n\\n已进行参数初始化操作\\n\').pack(side=tk.LEFT)
    row4 = tk.Frame(self)
    row4.pack(fill="x")
    assist.write_txt_data([\'1\',\'2\',\'2\',\'2\',\'1\',\'3\',\'1\',\'2\',\'5\',\'5\',\'5\',\'0\'],self.data_set_path )
    
    
    tk.Button(row4, text="关闭窗口", command=self.cancel).pack(side=tk.RIGHT)
  def setup_UI(self):
    
    self.A1 = tk.StringVar()
    self.A2 = tk.StringVar()
    self.A3 = tk.StringVar()
    self.A0 = tk.IntVar()
    
    self.B1 = tk.StringVar()
    self.B2 = tk.StringVar()
    self.B3 = tk.StringVar()
    self.B0 = tk.IntVar()
    
    self.C0 = tk.StringVar()
    self.C1 = tk.StringVar()
    self.C2 = tk.StringVar()
    self.C3 = tk.StringVar()
    
    data = assist.read_txt_data(self.data_set_path)
    self.A1.set(data[1])
    self.A2.set(data[2])
    self.A3.set(data[3])
    d = int(data[0])
    self.A0.set(d)
    
    self.B1.set(data[5])
    self.B2.set(data[6])
    self.B3.set(data[7])
    d = int(data[4])
    self.B0.set(d)
    
    self.C0.set(data[8])
    self.C1.set(data[9])
    self.C2.set(data[10])
    self.C3.set(data[11])
    
    row1 = tk.Frame(self)
    row1.pack(fill="x")
    
    tk.Label(row1,text=\'程序A\',width=30 ).grid(row=0,column=0,columnspan=2)
    tk.Label(row1,text=\'逗号次数:\',width=15).grid(row=1,column=0)
    tk.Label(row1,text=\'句号次数:\',width=15).grid(row=2,column=0)
    tk.Label(row1,text=\'全部文本次数\',width=15).grid(row=3,column=0)
    tk.Label(row1,text=\'是否生成文本\',width=15).grid(row=4,column=0)
    
    tk.Entry(row1, textvariable=self.A1, width=13).grid(row=1,column=1)
    tk.Entry(row1, textvariable=self.A2, width=13).grid(row=2,column=1)
    tk.Entry(row1, textvariable=self.A3, width=13).grid(row=3,column=1)
    checkA = tk.Checkbutton(row1, text="播放", variable=self.A0)
    checkA.select()
    checkA.grid(column=1, row=4, sticky=tk.W)
    
    
    tk.Label(row1,text=\'程序B\',width=30 ).grid(row=0,column=2,columnspan=2)
    tk.Label(row1,text=\'逗号次数:\',width=15).grid(row=1,column=2)
    tk.Label(row1,text=\'句号次数:\',width=15).grid(row=2,column=2)
    tk.Label(row1,text=\'全部文本次数\',width=15).grid(row=3,column=2)
    tk.Label(row1,text=\'是否生成文本\',width=15).grid(row=4,column=2)
    
    tk.Entry(row1, textvariable=self.B1, width=13).grid(row=1,column=3)
    tk.Entry(row1, textvariable=self.B2, width=13).grid(row=2,column=3)
    tk.Entry(row1, textvariable=self.B3, width=13).grid(row=3,column=3)
    checkB = tk.Checkbutton(row1, text="播放", variable=self.B0)
    checkB.select()
    checkB.grid(column=3, row=4, sticky=tk.W)
    
    self.A0.set(int(data[0]))
    self.B0.set(int(data[4]))
    
    #这个是音频合成设置界面的
    tk.Label(row1,text=\'音频合成参数设置\',width=30 ).grid(row=0,column=4,columnspan=2)
    tk.Label(row1,text=\'语速:\',width=15).grid(row=1,column=4)
    tk.Label(row1,text=\'音调:\',width=15).grid(row=2,column=4)
    tk.Label(row1,text=\'音量\',width=15).grid(row=3,column=4)
    tk.Label(row1,text=\'发音人\',width=15).grid(row=4,column=4)
    
    tk.Entry(row1, textvariable=self.C0, width=13).grid(row=1,column=5)
    tk.Entry(row1, textvariable=self.C1, width=13).grid(row=2,column=5)
    tk.Entry(row1, textvariable=self.C2, width=13).grid(row=3,column=5)
    tk.Entry(row1, textvariable=self.C3, width=13).grid(row=4,column=5)
    
    tk.Button(row1, text="取消", command=self.cancel, width=13).grid(row=9,column=0)
    tk.Button(row1, text="确定", command=self.ok, width=13).grid(row=9,column=4)
    tk.Button(row1, text="重置", command=self.reset, width=13).grid(row=9,column=5)
    
  def ok(self):
    self.userinfo_A = [self.A0.get(), self.A1.get(),self.A2.get(),self.A3.get()] # 设置数据
    self.userinfo_B = [self.B0.get(), self.B1.get(),self.B2.get(),self.B3.get()] # 设置数据
    self.userinfo_C = [self.C0.get(), self.C1.get(),self.C2.get(),self.C3.get()]
    
    self.userinfo_A[0] = str(self.userinfo_A[0])
    self.userinfo_B[3] = str(self.userinfo_B[3])
    
    
    data = self.userinfo_A + self.userinfo_B+self.userinfo_C
    # print(data)
    assist.write_txt_data(data,os.path.dirname(os.path.realpath(__file__))+"\\\\data_set.txt")
    self.destroy() # 销毁窗口
  def cancel(self):
    self.userinfo = None # 空!
    self.destroy()
  
  def reset(self):
    assist.write_txt_data([\'1\',\'2\',\'2\',\'2\',\'1\',\'3\',\'1\',\'2\',\'5\',\'5\',\'5\',\'0\'],self.data_set_path )
    self.destroy()
    
################################
class main_code(tk.Tk):
  def __init__(self):
    super().__init__()
    self.audio_path = ""
    self.run = True
    self.vartext = tk.StringVar()
    
    self.txt=\'\'
    networkisok=isConnected()
    if networkisok:
      self.mark_text = \'网络连接正常!\\n欢迎使用!\'
    else:
      self.mark_text = \'无网络连接!\\n请确认网络连接是否完好\\n无法连接百度云服务器\'
    
    self.insert_text = \'\'
    self.vartext.set(self.mark_text)
    self.audio_path = ""
    self.title(\'音频标记\')
    self.setupUI()
    
  def setupUI(self):
    
    row4 = tk.Frame(self)
    row4.pack(fill="x")
    
    tk.Button(row4,text=\'导入文本\',width=27,command=self.select).grid(row=0,column=0)
    tk.Button(row4,text=\'设置文本\',width=27,command=self.setup_config).grid(row=0,column=1)
    tk.Button(row4,text=\'生成文本\',width=27,command=lambda:self.lay(entry1.get())).grid(row=0,column=2)
    
    label = tk.Label(row4,height=3,textvariable=self.vartext, bg=\'white\', font=(\'黑体\', 10), anchor=\'w\').grid(row=1,column=0,columnspan=3)
    
    entry1 = tk.Entry(row4,width=85)
    entry1.grid(row=2,column=0,columnspan=3)

  def select(self):   #获取audio_path和audio_len
    self.audio_path = tk.filedialog.askopenfilename(initialdir = \'C:\\\\Users\\\\Administrator\\\\Desktop\',filetypes=( ("txt or docx files", "*.txt;*.docx"),("All files", "*.*")))
    if self.audio_path=="":
      self.vartext.set(\'未导入任何文件\')
    else:
      p,f=os.path.split(self.audio_path)
      path = os.path.dirname(os.path.realpath(__file__)) + \'\\\\data\\\\\' + str(os.path.splitext(f)[0])
      # print(path)
      isExists = os.path.exists(path)
      if not isExists:  #如果不存在这个目录
        os.makedirs(path)  #创建一个新的路径
        shutil.copyfile(self.audio_path,path+\'\\\\\'+f)
        
      if str(os.path.splitext(f)[1])==\'.txt\':
        file_object = open(self.audio_path,\'r\', encoding=\'UTF-8\')
        try:
          file_context = file_object.read()
        finally:
          file_object.close()
        self.txt=file_context
        self.vartext.set(\'已导入txt文本\')
      # print(1)
      elif str(os.path.splitext(f)[1])==\'.docx\':
        import docx
        document = docx.Document(self.audio_path)
        txt=\'\'
        for paragraph in document.paragraphs:
          txt+=paragraph.text
        self.txt=txt
        self.vartext.set(\'已导入docx文本\')
      
      self.audio_path=path+\'\\\\\'+f#更新数据路径
      # print(self.audio_path)
    
  def lay(self,txt=\'\'):
    mark_set=assist.read_txt_data(os.path.dirname(os.path.realpath(__file__))+"\\\\data_set.txt")
    # print(mark_set)
    
    for i in range(len(mark_set)):
      mark_set[i]=int(mark_set[i])
    
    # mark_set=[1,1,0,1,0,0,1,4]
    if txt==\'\':
      if self.audio_path!=\'\':
        p,f=os.path.split(self.audio_path)
        # print(p)
        if mark_set[0]==1:
          # code_a(p+\'\\\\\'+str(os.path.splitext(f)[0])+\'_A\'+\'.txt\',self.txt,mark_set[1:4])
          code_a(p+\'\\\\\'+str(os.path.splitext(f)[0])+\'_A\',self.txt,mark_set[1:4],mark_set[8:12])
          # self.vartext.set(\'已生成文本标记文件A\')
        if mark_set[4]==1:
          # code_b(p+\'\\\\\'+str(os.path.splitext(f)[0])+\'_B\'+\'.txt\',self.txt,mark_set[5:])
          code_b(p+\'\\\\\'+str(os.path.splitext(f)[0])+\'_B\',self.txt,mark_set[5:],mark_set[8:12])
          # self.vartext.set(\'已生成文本标记文件B\')
        if mark_set[0]==1 and mark_set[4]==1:
          self.vartext.set(\'已生成文本标记文件A、B和相应的音频\')
        elif mark_set[0]==1 and mark_set[4]==0:
          self.vartext.set(\'已生成文本标记文件A和相应的音频\')
        elif mark_set[0]==0 and mark_set[4]==1:
          self.vartext.set(\'已生成文本标记文件B和相应的音频\')
        elif mark_set[0]==0 and mark_set[4]==0:
          self.vartext.set(\'设置中未确定生成何种文件\\n请在设置文本中调整参数!\')
      else:
        self.vartext.set(\'未导入任何文件\')
    else:
      # print(txt)
      if mark_set[0]==1:
        code_a(\'\',txt,mark_set[1:4])
        self.vartext.set(\'已将A程序处理后的文字复制到剪切板\\n直接粘贴即可\\n如果想使用B程序处理请在设置中将程序A取消,并勾选程序B\')
      elif mark_set[4]==1:
        code_b(\'\',txt,mark_set[5:])
        self.vartext.set(\'已将B程序处理后的文字复制到剪切板\\n直接粘贴即可\\n如果想使用A程序处理请在设置中将程序B取消,并勾选程序A\')
    
    
  def setup_config(self):
    res = self.ask_userinfo()
    if res is None: return
  def ask_userinfo(self):
    inputDialog = out_setconfig()
    self.wait_window(inputDialog) # 这一句很重要!!!

if __name__ == \'__main__\':
  app 以上是关于文本转音频(百度语音合成api)(python)的主要内容,如果未能解决你的问题,请参考以下文章

基于百度云实现微信服务号内文字语音互转STT TTS

Android TextToSpeech TTS中文文本转语音(语音合成)

Android TextToSpeech TTS中文文本转语音(语音合成)

如何优雅的用Python玩转语音聊天机器人

文本转语音通过语音合成标记语言(SSML)改进合成 知识点详解

Android 百度语音合成 (含离线在线API合成方式,详细步骤+源码)