ISSN (Online) 2581-9429

IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 3, Issue 2, April 2023

Impact Factor: 7.301

Farmer’s Assistant using AI Voice Bot

Prof. D. R. Anekar1, Saurabh Suryavanshi2, Dnyanesh Auti3,
Praphulla Lokhande4, Aditya Deshmukh5
Assistant Professor, Department of Information Technology1
Students, Department of Information Technology2,3,4,5
Sinhgad Academy of Engineering, Pune, Maharashtra, India
[email protected], [email protected],
[email protected], [email protected]

Abstract: The agricultural sector contributed approximately 19% of country’s total Gross Domestic
Product and employ’s in an around 62% of the India’s population in 2020, making it an essential part of
the nation's economic development. This figure was previously around 15.41% until 2018, but the
introduction of cutting-edge technologies such as IoT, AI, and ML has helped to increase this number. A
system is created to assist farmers through a voicechat bot. The voicechat bot will address farmer queries
in various languages, the voicechat bot has been developed using Google translate and its dataset. This
application aims to help farmers enhance their farming practices and thus growth in their crop production.
By providing guidance on crop selection and pesticide usage, this technology is expected to lead to an
overall growth in the agriculture sector.

Keywords: Voice Chat Bot, Agriculture, And Gross Domestic Production.

I. INTRODUCTION
Till 2050, the global population is predicted to exceed 9 billion people, and this will require a 70 percent increase in
agricultural production to meet needs. However, current agricultural performing methods require high energy inputs,
and the market demands high-quality food. The scarcity of labor, rising labor costs, crop failures due to pests, drought,
climate changes, soil fertility loss, and fluctuating market prices for agricultural products have all negatively affected
the social and economic status of this crucial population. On the other side, the growing population has resulted in
higher demand of food grains, leading to the system provides support in three native languages: English, Hindi, and
Marathi.
It consists of two modules: Base Input and Registration of Profile. For farming-related consultations, a recommendation
algorithm is utilized, which leverages natural language processing and neural networks to identify and address farmer
queries. Inflation in agricultural product prices. In 2015-16, 95.18% of people in India were protected by the National
Foods Securities Act 2013. The agricultural sector in India faces significant challenges and needs to expand more
rapidly to meet the rising demand of the population. Despite the support for research and development, as well as
technology, this industry has not seen significant growth. However, with the emergence of new technologies such as
smartphones, it is easier to provide technology to farmers through mobile phones. Several initiatives have been
launched with the latest technological advancements in mind, including the Indian government's Kisan Call Center
(KCC) program, which provides information about the government's interactions with farmers. The KCC dataset is a
resource that is available to the public for research and analysis. To overcome the language barrier and provide support
to local users in their native language, the system will include a tool that can be used in localized languages, such as
English, Hindi, and Marathi .

II. LITERATURE SURVEY

S. Kanaga et al., (2021) [1] have proposed the use of an interactive chatbot in e-agriculture to enable farmers to sell
their products directly to consumers without the involvement of middlemen. The proposed chatbot is accessible to
consumers on any social media platform, allowing farmers to communicate via a network (like Facebook Messenger)
and cultivate a specific crop for a fee. Through the aid of a smart system, consumers can hire farmers who can supply
Copyright to IJARSCT DOI: 10.48175/IJARSCT-9121 224
www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 3, Issue 2, April 2023

Impact Factor: 7.301

the food and provide all the necessary inputs for farming, resulting in a fair and profitable price for both buyers and
sellers.
Anchal Jaiswalet al., (2020) [2] The AgriBot is a portable and intelligent system that assists farmers in choosing the
best crops and farming strategies based on the current weather, soil and geographical characteristics of the designated
area. The lack of a single resource that can address all of the farmers' questions about market prices, storage facilities,
seeds, fertilizers, and government schemes is a challenge. To address this issue, the study proposes the use of a chatbot
that utilizes natural language
processing methodology. The chatbot enables farmers to enter agricultural-related
agricultural related questions in audio format,
simplifying farmer contact. In case the system fails to respond to any of the specified queries, the request is sent to
assistance lines.
Vijay S. Rajpurohitet al., (2019) [3]] This paper presents a survey on chat-bot bot systems for the agriculture domain.
Chat-bot
bot systems use natural language processing to understand user needs and provide appropriate responses.
Agriculture is a vital sector for national growth, but many farmers lack knowledge of modern farming methods and
technologies. Machine learning techniques have been applied to enable chat-bots
chat to provide
vide meaningful responses to
agricultural queries. This survey examines expert systems and cutting-edge
cutting edge machine learning methods used for question
and answer systems in agriculture. Chat-bot
Chat bot systems enable farmers to easily make inquiries and receive speci specific
answers to their questions.
Dheeraj Singh Chaudharyet et al., (2020) [4] Agribot is a conversational chatbot designed for agricultural applications,
which not only provides answers to frequently asked questions but also focuses on crop disease detection aand weather
forecasting. The system is based on an end-to-end
end trainable sequence-to-sequence
sequence learning model, which is designed to
achieve conversational task-oriented
oriented communication with minimal assumptions on sequence structure.

II. PROPOSED SYSTEM ARCHITECTURE

ARCHITE

Fig. 1: Proposed system architecture

Over the past few years, we have collected data from a government website with information on all states in India. As
part of our project, we have developed voices and Chabot’s.Shachi Mall et al. [10], Xu Wu et al. [11], N. Shelke et. al.
[12] and S. L. Bangare et. al. [13-16],
16], Ajay S. Ladkat et al. [17] and V. Durga Prasad Jasti et. al. [18] have shown
different methods for machine learning.

2.1 Data Analysis

Data analysis is the first step in analysing and responding
responding to farmers' productions. In southern India, irrigation is an
important element of food security. It is the application of water to the ground to grow. Various studies have been
carried out and the crops are analyzed and recommendations are tabulated
tabulated based on parameters such as crops, yield,
productivity, agro climate, and soil type we will make recommendations for crops and fertilizers.

Copyright to IJARSCT DOI: 10.48175/IJARSCT-9121 225

www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 3, Issue 2, April 2023

Impact Factor: 7.301

A. Convolutional Layer
The main function of the convolution layer is to perform pattern recognition by performing a series of convolution
operations on its input. During forward propagation, this layer convolves each input with a filter or kernel, producing an
output that exposes the temporal moments that affect the classification.

B. Pooling Layer
The grouping layer divides the input dimension into a nonlinear layer to ensure that the most relevant information is
retained. In order to perform its down sampling function, a clustering layer must be configured with a clustering
window and a clustering step. Typically, these variables are set to the same value to avoid overlap, and the window
moves through the input data and selects a segment of input values to apply to the grouping function. This down-
sampling process is performed by the clustering layer.
 Max Pooling –The output of a pooling layer is determined by choosing the maximum value present in the
pooling window.
 Average Pooling – Unlike the previous statement, the output of a pooling layer is calculated as the average of
the values in the pooling window. An illustration of this function applied to a 1D input with both pooling
window and pooling step set to 2 can be seen in Figure 1-

C. Flatten layer
The flat layer is responsible for concatenating each intermediate feature map of the final convolution block, reducing
the 2-D space corresponding to the output dimension of the convolution part to a 1- space. D which can be entered in
the classification part of the network.
 Kolomgewys – C = t 0 [x0], t 1 [x0], t 2 [x0], ......., t n−1 [xd−1] ,
 Rygewys – C = t 0 [x0], t 0 [x1], t 0 [x2], ......., tn−1 [xd−1]

D. Fully-Connected Layers
Fully Connected Layer After the convolutional layer extracts relevant features from the input data, the fully connected
(FC) layer, denoted λ, combines each neuron in a way that facilitates accurate and efficient classification of the input .
FC layers can be compared to multi-layer perceptron (MLP) networks, which operate on the output of the previous
layer.

2.2 Voice Bots

Figure 2 illustrates the process in which a farmer provides voice input, which is then fed into a voice-to-text synthesizer
that utilizes Google Translator. The synthesizer converts the farmer's voice into text, which is then inputted into an

Copyright to IJARSCT DOI: 10.48175/IJARSCT-9121 226

www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 3, Issue 2, April 2023

Impact Factor: 7.301

information retrieval
etrieval block that searches for answers of the query by using a search Engine such as Google. The result
of the search is then converted back into the voice format by a Text-To-Voice
Text synthesizer.

Fig-02 Block Diagram of Voice Bot

Voice to Text Synthesizer:

 Enable for The Ambient Noise Adjustment: Due to the variance in ambient noise levels, it is essential to
allow the software a few seconds to Adjust the recording the energy threshold to match the external noise
level.
 Speech to the Text Translation: lation: Speech to text translation: Google voice recognition is used to convert
speech to text. This service requires an active Internet connection to operate. While some offline recognition
systems (such as Pocket Sphinx) have a long setup process and require req uire you to install lots of dependencies,
Google Speech Recognition is one of the easiest to use.
 Google Translation: Google Translator is a python library that utilized the Google Translator API to translate
text from regional languages into the English. It is a free and unlimited library that relies on the Google
Translates AJAX API to perform functions such as identification and translation. The library is compatible
with Python 3.6 and higher versions, and is known for its speed and reliability, leverag
leveraging the same servers as
translate.google.com. Additionally, it features auto language detection capabilities.
 Voice Repeater (pysttsx3): After the text has been translated, it is spoken aloud to confirm that the text
spoken by the farmer has been accurately
accurately captured. This is achieved using the pyttsx3 library, which is a
Python-based Text-To-Speech
Speech conversion Library. Unlike other libraries, it works offline and is compatible
with both Python 2,3. To obtain the references to a pyttsx3 Engine instances, an aapplication must call the
pyttsx3.init () Factories function.
 Information Retrieval (Using google search, web browser): The Google Search Python library can be used
to search for information on Google. This library uses queries and Beautiful Soup4 to retrie retrieve search results.
By default, Google returns ten search results. The Web Browser module can be used to open web pages in the
default browser. To use this module, you simply need to import it and call its open() function with a URL.
After using Google Translate
slate to translate the web page into the farmer's local language, the translated text can
be fed into a text-to-speech
speech synthesizer. This will allow the computer to speak the translated text in the
farmer's local language.

2.3 Suggestion Bot

 Crops Suggestions: Based on location and weather information provided by the farmer, crop suggestions are
made. The farmer provides the location where they plan to cultivate the crops as Input. This input is processed
using an algorithm that leverages to the services of the openweathermap.org to obtain weathers weathers-related
information for the specified location.
 Open weather map: Open Weather Map is a service that provides weather data for web services and mobile
applications. The service offers current weather data, forecasts,
forecasts, and historical data through an API with JSON,
XML, and HTML endpoints. There is a free usage level available, but users must obtain an API key by
creating an account on the Open Weather Map website. The API provides information such as current weathe weather
conditions, extended forecasts, and graphical maps depicting cloud cover, wind speed, pressure, and
precipitation.

Copyright to IJARSCT DOI: 10.48175/IJARSCT-9121 227

www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 3, Issue 2, April 2023

Impact Factor: 7.301

 Request and JSON: The algorithm uses the request module to fetch the data from the website and processes it
with help of the JSON Module. The The requests module is used to send the HTTP requests using Python, while
the JSON module is built-in in package in Python that enables working with JSON data. Based on the weather,
soil type, and region provided by the farmer, the algorithm suggests the type of crop that would be suitable for
cultivation. The suggestions are determined by referring to Table 2, which lists the suitable crops based on
climate, location, and the soil type.
 Fertilizer Suggestion Based On Nutrition: This algorithm utilizes machine learning decision tree
classification to suggest the appropriate fertilizer for a crop based on the results of a soil test. By analysing the
nutritional composition of the soil, the algorithm can determine the optimal fertilizer that will promote a good
crop yield.

III. RESULT
The Voice Bot enables farmers to ask questions using their regional language through voice input. Once the input is
provided, it is converted into text, and the bot retrieves the appropriate information fertilizers. In Fig.03, the farmer
selects their mother tongue and provides voice input to the bot, which retrieves information from websites to answer the
query. Fig.04 displays the results obtained by the bot. The suggestion bot is specifically designed to aid farmers in
selecting crops based on weather and soil conditions. It can also provide fertilizer recommendations based on soil test
ratings. Upon using the suggestion bot, the final output consists of recommended crops along with suggested fertilizers.

Fig 03. Registration and Login

Fig--04 Results of Crop and Fertilizer Suggestions

Performance Check with Monte Carlo Cross Validation: The entire Ground Truth (GT) dataset is partitioned into k
subsets or "folds". To achieve this, the data is randomly shuffled and 80% of the GT is selected for training, while the
remaining 20% is reserved for testing. The shuffling process is repeated multiple times to obtain several accuracy
values, and the average accuracy is calculated. For instance, if the total GT is 25 and the shuffling process is repeated 5
times, resulting in the following accuracies:
accuracies After shuffling the data five times, the accuracy values were as follows:
Shuffle 1 Accuracy = 64 (16+9), Shuffle 2 Accuracy = 72 (18+7), Shuffle 3 Accuracy = 70 (17+8), Shuffle 4 Accuracy
= 80 (20+5), Shuffle 5 Accuracy = 70 (17+8). The average accuracy is calculated as 71.2, which is obtained by
summing up correct predictions (17) and incorrect predictions (8) and dividing the result by the total number of

Copyright to IJARSCT DOI: 10.48175/IJARSCT-9121 228

www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 3, Issue 2, April 2023

Impact Factor: 7.301

instances (25).The NLU system recognized 25 user inputs, with 17 being predicted correctly and 8 being predicted
incorrectly. The precision is calculated as ~0.712 (17/ (17+8)), which is the number of correct predictions for the user
input divided by the total number of predictions for the user input. The accuracy of the system is approximately 71%,
which is lower than that of some other text bots, but compared to a voice bot, it is considered very efficient.

"Suggestion Bot: Crops and fertilizers suggestion based on the Weather, Location and Soil.

Fig. 05. Result of the Crop And Fertilizer Suggestion

IV. CONCLUSION
The implementation of an AI voice bot system as a farmer's assistant can provide significant assistance to the
agricultural community. This application offers farmers access to relevant agriculture related data, and general
information such as weather forecasts, recommended crops for planting, and suitable fertilizers to use. The system is
designed to be multilingual and includes a voiceover feature, as well as entity extraction for farmer queries. This
enables farmers from different regions and with different language preferences to ask queries and receive
responses/answers in their regional language. The voice bot system can also suggest crops, pesticides depending on the
climate and soil type, thus making modern farming technology accessible to a vast community of farmers. As a next
step, the system could be expanded to identify crop diseases and provide remedies for them. Additionally, the system
could suggest better crops and better time for their harvest based on market rates, climatic conditions, and land
conditions

REFERENCES
[1]. Kiruthika, Usha & Subramanian, Kanaga Suba & Balaji, V. & Raman, C.J.. (2020). “E-Agriculture for Direct
Marketing of Food Crops Using Chatbots”, 1-4. 10.1109/ICPECTS49113.2020.9337024.
[2]. D. Sawant, A. Jaiswal, J. Singh and P. Shah, “AgriBot - An intelligent interactive interface to assist farmers in
agricultural activities”, 2019 IEEE Bombay Section Signature Conference (IBSSC), Mumbai, India, 2019, pp.
1-6, doi: 10.1109/IBSSC47189.2019.8973066.
[3]. P. Y. Niranjan, V. S. Rajpurohit and R. Malgi, "A Survey on Chat-Bot system for Agriculture Domain," 2019
1st International Conference on Advances in Information Technology (ICAIT), Chikmagalur, India, 2019, pp.
99-103, doi: 10.1109/ICAIT47043.2019.8987429.
[4]. B. Arora, D. S. Chaudhary, M. Satsangi, M. Yadav, L. Singh and P. S. Sudhish, "Agribot: A Natural
Language Generative Neural Networks Engine for Agricultural Applications," 2020 International Conference
on Contemporary Computing and Applications (IC3A), Lucknow, India, 2020, pp. 28-33, doi:
10.1109/IC3A48958.2020.233263.
[5]. G. M. D'silva, S. Thakare, S. More and J. Kuriakose, "Real world smart chatbot for customer care using a
software as a service (SaaS) architecture," 2017 International Conference on I-SMAC (IoT in Social, Mobile,
Analytics and Cloud) (I-SMAC), Palladam, India, 2017, pp. 658-664, doi: 10.1109/I-SMAC.2017.8058261.

Copyright to IJARSCT DOI: 10.48175/IJARSCT-9121 229

www.ijarsct.co.in
 ISSN (Online) 2581-9429
IJARSCT
International Journal of Advanced Research in Science, Communication and Technology (IJARSCT)

Volume 3, Issue 2, April 2023

Impact Factor: 7.301

[6]. Aleksandra Przegalinska, Leon Ciechanowski, Anna Stroz, Peter Gloor, Grzegorz Mazurek, “In bot we trust:
A new methodology of chatbot performance measures”, Business Horizons, Volume 62, Issue 6,2019, Pages
785-797, ISSN 0007-6813, https://doi.org/10.1016/j.bushor.2019.08.005.
[7]. Xiaolan Fu, Shaheen Akter (2012) In research paper presented at the International Conference on Agriculture
Economist discussion on Quality and Speed of Agriculture Extension Evidence from the Rural e-services
Project in India’ International Conference on Agriculture Economist, 2012, issue no 2, (1-32.)
[8]. Mostaco, Gustavo Marques, IcaroRamires Costa De Souza, Leonardo Barreto Campos, and Carlos Eduardo
Cugnasca. (2018) Agro-nomo-Bot a smart que-ans Chatbot applied to agriculture s networks.” In 14th
international conference on precision agriculture, vol. 24,(pp. 1-13)
[9]. A.B Abacha, P Zweigenbaum”(September 2015)MEANS: A medical que-ans system combining Natrual
Language Processing techniques and semantic web technologies”, Information Processing and Management,
vol 51, Issue. 5, (pp 570- 594)
[10]. Shachi Mall, Ashutosh Srivastava, Bireshwar Dass Mazumdar, Manmohan Mishra, Sunil L. Bangare, A.
Deepak, “Implementation of machine learning techniques for disease diagnosis”, Materials Today:
Proceedings, Volume 51, Part 8, 2022, Pages 2198-2201, ISSN 2214-7853,
https://doi.org/10.1016/j.matpr.2021.11.274.
[11]. Xu Wu, Dezhi Wei, Bharati P. Vasgi, Ahmed Kareem Oleiwi, Sunil L. Bangare, Evans Asenso, "Research on
Network Security Situational Awareness Based on Crawler Algorithm", Security and Communication
Networks, vol. 2022, Article ID 3639174, 9 pages, 2022. https://doi.org/10.1155/2022/3639174
[12]. N. Shelke, S. Chaudhury, S. Chakrabarti, S. L. Bangare et al. “An efficient way of text-based emotion
analysis from social media using LRA-DNN”, Neuroscience Informatics, Volume 2, Issue 3, September
2022, 100048, ISSN 2772-5286, https://doi.org/10.1016/j.neuri.2022.100048
[13]. S. L. Bangare, G. Pradeepini and S. T. Patil, "Brain tumor classification using mixed method approach," 2017
International Conference on Information Communication and Embedded Systems (ICICES), Chennai, India,
2017, pp. 1-4, doi: 10.1109/ICICES.2017.8070748
[14]. S. L. Bangare, G. Pradeepini, S. T. Patil, “Implementation for brain tumor detection and three dimensional
visualization model development for reconstruction”, ARPN Journal of Engineering and Applied Sciences
(ARPN JEAS), Vol.13, Issue.2, ISSN 1819-6608, pp.467-473. 20/1/2018 http://www.arpnjournals.org/jeas/
research_papers/rp_2018/jeas_0118_6691.pdf
[15]. S. L. Bangare, “Classification of optimal brain tissue using dynamic region growing and fuzzy min-max
neural network in brain magnetic resonance images”, Neuroscience Informatics, Volume 2, Issue 3,
September 2022,100019, ISSN 2772-5286, https://doi.org/10.1016/j.neuri.2021.100019
[16]. Sunil L. Bangare, Deepali Virmani, Girija Rani Karetla, Pankaj Chaudhary, Harveen Kaur, Syed Nisar
Hussain Bukhari, Shahajan Miah, "Forecasting the Applied Deep Learning Tools in Enhancing Food Quality
for Heart Related Diseases Effectively: A Study Using Structural Equation Model Analysis", Journal of Food
Quality, vol. 2022, Article ID 6987569, 8 pages, 2022. https://doi.org/10.1155/2022/6987569
[17]. Ajay S. Ladkat, Sunil L. Bangare, Vishal Jagota, Sumaya Sanober, Shehab Mohamed Beram, Kantilal Rane,
Bhupesh Kumar Singh, "Deep Neural Network-Based Novel Mathematical Model for 3D Brain Tumor
Segmentation", Computational Intelligence and Neuroscience, vol. 2022, Article ID 4271711, 8 pages, 2022.
https://doi.org/10.1155/2022/4271711
[18]. V. Durga Prasad Jasti, Enagandula Prasad, Manish Sawale, ShivlalMewada, Manoj L. Bangare, Pushpa M.
Bangare, Sunil L. Bangare, F. Sammy, "Image Processing and Machine Learning-Based Classification and
Detection of Liver Tumor",BioMed Research International, vol. 2022, Article ID 3398156, 7 pages, 2022.
https://doi.org/10.1155/2022/3398156.

Copyright to IJARSCT DOI: 10.48175/IJARSCT-9121 230

www.ijarsct.co.in