A Simple Voice-Based Chatbot For Hindi Language

 A Simple Voice-Based Chatbot For Hindi Language

A Voice-Based Chatbot acts like a voice assistant to accept your vocal queries, process them and produce the vocal answer to satisfy the query. 

In this article, we will create a simple voice-based chatbot for Hindi language. The chatbot will accept the Hindi queries through vocal commands, will process them with text-matching principle and will revert with appropriate vocal answer in Hindi.

To do this following Python libraries will be helpful.         

1. speech_recognition: Speech recognition is a machine's ability to listen to spoken words and identify them. You can then convert the spoken words into text, make a query or give a reply. The code uses this library to recognize the input Hindi query. (installation: pip install SpeechRecognition)
2. translate: Translate is a simple but powerful translation tool written in python with with support for multiple translation providers. The code uses this library to translate the input Hindi query into English for text-matching. (Installation: pip install translate)
3. gTTS:  A gTTS (Google Text-to-Speech), is a Python library to interface with Google Translate's text-to-speech API. It can process the .mp3 input file. (Installation: pip install gTTS). The code uses this library to translate the response in Hindi. You can reuse the above mentioned library 'translate' instead. However the code uses this specific library to explore to the geeks, the another available option. 
4. pygame: It is a free and open-source cross-platform library for the development of multimedia applications including audio and video. (Installation: pip install pygame). The code uses this library to save and play the translated response in Hindi.

The Code is explained below.

Step1 : Import the libraries you need.

Python3

# import required libs

import random

import speech_recognition as sr

from translate import Translator

from gtts import gTTS

from datetime import datetime

from pygame import mixer

 

Step 2:  Receive A Vocal Query in Hindi: To do this we create and initialize object of speech_recognition. It helps to recognize the audio query  obtained from the device-microphone as the source. It recognizes the spoken Hindi words using the recognize_google().     

Python3

# the function is coded to take Hindi input queries and recognize them

def Receive_In_Hindi():

    r = sr.Recognizer()

    with sr.Microphone() as source:

        print('Listening')

        r.pause_threshold = 0.7

        audio = r.listen(source)

        try:

            print("Recognizing")

            Query = r.recognize_google(audio, language='hi-In')

            Answer_In_Hindi(Query)

            # handling the exception, so that assistant can

            # ask for telling again the command

        except Exception as e:

            print(e)

            print("Didn't Get That ! Please say again..")

            return "None"

        return Query

 

Step 3:  Translate the vocal query to English and process response : This is required as our code does 'text-matching' for processing the query. The input vocal Hindi query needs to be translated into English first. The Translator() helps in this task. Then a response is generated  in English text format. 

Python3

def Answer_In_Hindi(Question):

        print("Your Query:",Question)

        #translated_text = GoogleTranslator(source='hi', target='en').translate(Question)

        translator= Translator(from_lang="hindi",to_lang="english")

        translation = translator.translate(Question)

   

        if "today" in translation:  # check if the query is about today

            answer_text="today is "+ datetime.today().strftime('%A')

            translator= Translator(from_lang="english",to_lang="hindi")

            translation = translator.translate(answer_text)

            Process_For_Hindi(translation)

       

        if "age" in translation: # check if the query is about age

            answer_text="I am 22 years old"

            translator= Translator(from_lang="english",to_lang="hindi")

            translation = translator.translate(answer_text)

            Process_For_Hindi(translation)

 

Step 4: Process the English-textual response into Hindi Audio:  As we need the final response in Hindi-audio format, we need to convert the processed textual response into Hindi audio. The gTTS() does this task. Since all responses need to be saved, we generate a random text file name and save the audio responses.

Python3

def Process_For_Hindi(translation):

    print("Chatbot Response:"+translation)

    answer_text=str(translation) #translate the English response into Hindi and convert it into audio file

    myobj = gTTS(text=answer_text, lang='hi', slow=False)

    file_name="chat_aswer"+str(random.randint(0,100))+".mp3"

    myobj.save(file_name)

    mixer.init()

    mixer.music.load(file_name)

    mixer.music.play()   #play the audio response

 

Output: 

The Input Query and the Response

 

The input should be provided in Hindi to the microphone of the device. When the query is processed, the code plays the Hindi audio file generated as the response of the query.