使用Python创建实时语音翻译器

在本文中，我们将使用Python创建一个实时语音翻译器。

需要模块

• playsound：该模块用于在Python中播放声音

pip install playsound

• 语音识别模块：它是一个库， Python可以借助它来识别给定的命令。我们必须使用 pip 进行语音识别。

pip install SpeechRecognition

• googletrans ： Googletrans 是一个免费且无限制的Python库，实现了 Google Translate API

pip install googletrans

• gTTs： gTTS API 支持多种语言，包括英语、印地语、泰米尔语、法语、德语等等。

pip install gTTs
pip install gTTS-token

一个实时语音翻译器，可以翻译语音输入并给出翻译后的语音输出。它是使用谷歌的 googleTrans API 和Python的 Speech_recognition 库创建的。它将文本从一种语言转换为另一种语言并保存其 mp3 录制文件。然后使用 playsound 模块播放生成的 mp3 文件，然后使用 os 模块删除生成的 mp3 文件。

逐步实施

第 1 步：导入必要的模块

Python3

# Importing necessary modules required 
from playsound import playsound
import speech_recognition as sr 
from googletrans import Translator 
from gtts import gTTS 
import os

第 2 步：所有语言与其代码映射的元组

Python3

dic=('afrikaans', 'af', 'albanian', 'sq', 'amharic', 'am', 
     'arabic', 'ar', 'armenian', 'hy', 'azerbaijani', 'az',
 'basque', 'eu', 'belarusian', 'be', 'bengali', 'bn', 'bosnian',
     'bs', 'bulgarian', 'bg', 'catalan', 'ca',
  'cebuano', 'ceb', 'chichewa', 'ny', 'chinese (simplified)',
     'zh-cn', 'chinese (traditional)', 'zh-tw',
  'corsican', 'co', 'croatian', 'hr', 'czech', 'cs', 'danish',
     'da', 'dutch', 'nl', 'english', 'en', 'esperanto',
  'eo', 'estonian', 'et', 'filipino', 'tl', 'finnish', 'fi', 
     'french', 'fr', 'frisian', 'fy', 'galician', 'gl',
  'georgian', 'ka', 'german', 'de', 'greek', 'el', 'gujarati', 
     'gu', 'haitian creole', 'ht', 'hausa', 'ha', 
  'hawaiian', 'haw', 'hebrew', 'he', 'hindi', 'hi', 'hmong', 
     'hmn', 'hungarian', 'hu', 'icelandic', 'is', 'igbo',
  'ig', 'indonesian', 'id', 'irish', 'ga', 'italian', 'it', 
     'japanese', 'ja', 'javanese', 'jw', 'kannada', 'kn',
  'kazakh', 'kk', 'khmer', 'km', 'korean', 'ko', 'kurdish (kurmanji)',
     'ku', 'kyrgyz', 'ky', 'lao', 'lo', 
  'latin', 'la', 'latvian', 'lv', 'lithuanian', 'lt', 'luxembourgish',
     'lb', 'macedonian', 'mk', 'malagasy',
  'mg', 'malay', 'ms', 'malayalam', 'ml', 'maltese', 'mt', 'maori',
     'mi', 'marathi', 'mr', 'mongolian', 'mn',
  'myanmar (burmese)', 'my', 'nepali', 'ne', 'norwegian', 'no',
     'odia', 'or', 'pashto', 'ps', 'persian',
   'fa', 'polish', 'pl', 'portuguese', 'pt', 'punjabi', 'pa',
     'romanian', 'ro', 'russian', 'ru', 'samoan',
   'sm', 'scots gaelic', 'gd', 'serbian', 'sr', 'sesotho', 
     'st', 'shona', 'sn', 'sindhi', 'sd', 'sinhala',
   'si', 'slovak', 'sk', 'slovenian', 'sl', 'somali', 'so', 
     'spanish', 'es', 'sundanese', 'su', 
  'swahili', 'sw', 'swedish', 'sv', 'tajik', 'tg', 'tamil',
     'ta', 'telugu', 'te', 'thai', 'th', 'turkish', 'tr',
  'ukrainian', 'uk', 'urdu', 'ur', 'uyghur', 'ug', 'uzbek', 
     'uz', 'vietnamese', 'vi', 'welsh', 'cy', 'xhosa', 'xh',
  'yiddish', 'yi', 'yoruba', 'yo', 'zulu', 'zu')

第 3 步：从用户那里获取语音命令

Python3

# Capture Voice
# takes command through microphone
def takecommand():
    r = sr.Recognizer()
    with sr.Microphone() as source:
        print("listening.....")
        r.pause_threshold = 1
        audio = r.listen(source)
  
    try:
        print("Recognizing.....")
        query = r.recognize_google(audio, language='en-in')
        print(f"user said {query}\n")
    except Exception as e:
        print("say that again please.....")
        return "None"
    return query

第 4 步：从用户那里获取语音输入

Python3

# Taking voice input from the user
query = takecommand()
while (query == "None"):
    query = takecommand()

步骤 5：从用户输入目标语言，将用户输入与语言代码映射

Python3

def destination_language():
    print("Enter the language in which you want to convert \
    : Ex. Hindi , English , etc.")
    print()
  
    # Input destination language in which the user 
    # wants to translate
    to_lang = takecommand()
    while (to_lang == "None"):
        to_lang = takecommand()
    to_lang = to_lang.lower()
    return to_lang
  
to_lang = destination_language()
  
# Mapping it with the code
while (to_lang not in dic):
    print("Language in which you are trying to convert\
    is currently not available ,please input some other language")
    print()
    to_lang = destination_language()
  
to_lang = dic[dic.index(to_lang)+1]

第 6 步：调用翻译器

Python3

# invoking Translator
translator = Translator()

第 7 步：从 src 转换为 dest

Python3

# Translating from src to dest
text_to_translate = translator.translate(query, dest=to_lang)
text = text_to_translate.text

第 8 步：保存翻译文件并在播放后删除它们

Python3

# Using Google-Text-to-Speech ie, gTTS() method
# to speak the translated text into the
# destination language which is stored in to_lang.
# Also, we have given 3rd argument as False because
# by default it speaks very slowly
speak = gTTS(text=text, lang=to_lang, slow=False)
  
# Using save() method to save the translated
# speech in capture_voice.mp3
speak.save("captured_voice.mp3")
  
# Using OS module to run the translated voice.
playsound('captured_voice.mp3')
os.remove('captured_voice.mp3')
print(text)

下面是完整的实现：

Python3

# Importing necessary modules required
from playsound import playsound
import speech_recognition as sr
from googletrans import Translator
from gtts import gTTS
import os
flag = 0
  
# A tuple containing all the language and
# codes of the language will be detcted
dic = ('afrikaans', 'af', 'albanian', 'sq', 
       'amharic', 'am', 'arabic', 'ar',
       'armenian', 'hy', 'azerbaijani', 'az', 
       'basque', 'eu', 'belarusian', 'be',
       'bengali', 'bn', 'bosnian', 'bs', 'bulgarian',
       'bg', 'catalan', 'ca', 'cebuano',
       'ceb', 'chichewa', 'ny', 'chinese (simplified)',
       'zh-cn', 'chinese (traditional)',
       'zh-tw', 'corsican', 'co', 'croatian', 'hr',
       'czech', 'cs', 'danish', 'da', 'dutch',
       'nl', 'english', 'en', 'esperanto', 'eo', 
       'estonian', 'et', 'filipino', 'tl', 'finnish',
       'fi', 'french', 'fr', 'frisian', 'fy', 'galician',
       'gl', 'georgian', 'ka', 'german',
       'de', 'greek', 'el', 'gujarati', 'gu',
       'haitian creole', 'ht', 'hausa', 'ha',
       'hawaiian', 'haw', 'hebrew', 'he', 'hindi',
       'hi', 'hmong', 'hmn', 'hungarian',
       'hu', 'icelandic', 'is', 'igbo', 'ig', 'indonesian', 
       'id', 'irish', 'ga', 'italian',
       'it', 'japanese', 'ja', 'javanese', 'jw',
       'kannada', 'kn', 'kazakh', 'kk', 'khmer',
       'km', 'korean', 'ko', 'kurdish (kurmanji)', 
       'ku', 'kyrgyz', 'ky', 'lao', 'lo',
       'latin', 'la', 'latvian', 'lv', 'lithuanian',
       'lt', 'luxembourgish', 'lb',
       'macedonian', 'mk', 'malagasy', 'mg', 'malay',
       'ms', 'malayalam', 'ml', 'maltese',
       'mt', 'maori', 'mi', 'marathi', 'mr', 'mongolian',
       'mn', 'myanmar (burmese)', 'my',
       'nepali', 'ne', 'norwegian', 'no', 'odia', 'or',
       'pashto', 'ps', 'persian', 'fa',
       'polish', 'pl', 'portuguese', 'pt', 'punjabi', 
       'pa', 'romanian', 'ro', 'russian',
       'ru', 'samoan', 'sm', 'scots gaelic', 'gd',
       'serbian', 'sr', 'sesotho', 'st',
       'shona', 'sn', 'sindhi', 'sd', 'sinhala', 'si',
       'slovak', 'sk', 'slovenian', 'sl',
       'somali', 'so', 'spanish', 'es', 'sundanese',
       'su', 'swahili', 'sw', 'swedish',
       'sv', 'tajik', 'tg', 'tamil', 'ta', 'telugu',
       'te', 'thai', 'th', 'turkish',
       'tr', 'ukrainian', 'uk', 'urdu', 'ur', 'uyghur',
       'ug', 'uzbek',  'uz',
       'vietnamese', 'vi', 'welsh', 'cy', 'xhosa', 'xh',
       'yiddish', 'yi', 'yoruba',
       'yo', 'zulu', 'zu')
  
  
# Capture Voice
# takes command through microphone
def takecommand():  
    r = sr.Recognizer()
    with sr.Microphone() as source:
        print("listening.....")
        r.pause_threshold = 1
        audio = r.listen(source)
  
    try:
        print("Recognizing.....")
        query = r.recognize_google(audio, language='en-in')
        print(f"The User said {query}\n")
    except Exception as e:
        print("say that again please.....")
        return "None"
    return query
  
  
# Input from user
# Make input to lowercase
query = takecommand()
while (query == "None"):
    query = takecommand()
  
  
def destination_language():
    print("Enter the language in which you\
    want to convert : Ex. Hindi , English , etc.")
    print()
      
    # Input destination language in
    # which the user wants to translate
    to_lang = takecommand()
    while (to_lang == "None"):
        to_lang = takecommand()
    to_lang = to_lang.lower()
    return to_lang
  
to_lang = destination_language()
  
# Mapping it with the code
while (to_lang not in dic):
    print("Language in which you are trying\
    to convert is currently not available ,\
    please input some other language")
    print()
    to_lang = destination_language()
  
to_lang = dic[dic.index(to_lang)+1]
  
  
# invoking Translator
translator = Translator()
  
  
# Translating from src to dest
text_to_translate = translator.translate(query, dest=to_lang)
  
text = text_to_translate.text
  
# Using Google-Text-to-Speech ie, gTTS() method
# to speak the translated text into the
# destination language which is stored in to_lang.
# Also, we have given 3rd argument as False because
# by default it speaks very slowly
speak = gTTS(text=text, lang=to_lang, slow=False)
  
# Using save() method to save the translated
# speech in capture_voice.mp3
speak.save("captured_voice.mp3")
  
# Using OS module to run the translated voice.
playsound('captured_voice.mp3')
os.remove('captured_voice.mp3')
  
# Printing Output
print(text)

输出：