ISSN:0975-9646

Mallamma V Reddy et al / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 14 (5) , 2023, 68-70

Machine translation tool for Hindi to English

Mallamma V Reddy1, Suraj Vasedar2
1,2
Department of Computer Science,
Rani Channamma University,Belagavi, Karnataka, India

ABSTRACT— The "Hindi to English Machine Translation 2. LITERATURE SURVEY

Tool" leverages natural language processing (NLP) as an In Natural Language Processing (NLP), MT serves as a
innovative approach to overcome linguistic barriers and conduit for inter language communication, MT uses
facilitate cross-cultural communication, this technology shines machine translation between two languages to solve
as a beacon of accessibility and understanding in a world that
is becoming more linked where language diversity can create
language ambiguity issues while maintaining meaning. As
impediments to good communication. The ideal of the MT systems advanced from rule based to corpus based
experimenter is to produce an intelligent, context aware methodologies, they did away with the need for linguistic
translation system that not only translates words from Hindi expertise, a never-ending list of NLP tasks, including
to English but also retains the subtleties, nuances, and cultural Named Entity Recognition, Speech tagging, Chunking,
sensitivities that are inherent in language. It produces a Word Sense Disambiguation, and the language diversity
translation tool that goes beyond literal conversions, our issue for Interlingua-based MT [1].
method blends cutting-edge neural machine translation analyzing MT systems within Indian languages and
(NMT) models with linguistic experience. We attempt to between Indian and international languages. It is simpler to
provide translations that accurately capture not just the words
but also the cultural nuances, context, and idiomatic
translate between language pairings with small structural
expressions. differences, such as Hindi and English, than between
language pairs with large structural differences, such as
Keywords— Cross-cultural communication, Hindi to English, Hindi and Punjabi [2].
Linguistic, Machine translation, Neural machine translation The long-term dependency problem and the sequence to
(NMT) sequence learning problem for source and target phrases of
various durations are both solved by the NMT system using
1. INTRODUCTION LSTM, the NMT system has outstanding context-analyzing
In a increasingly globalized world, communication knows no capabilities and enhances translation prediction [3].
borders. Language, once a formidable obstacle, can now be It is less common to use direct machine translation that is
easily overcome with machine translation tools. In this done at the word level. There is no intermediary
context, Hindi, a language spoken by millions of people, has representation used when translating words into the target
become an area of interest for the development and language [5].
improvement of machine translation systems. This research In Direct Machine Translation, there is a requirement for a
paper provides a brief overview of the current landscape of separate translator for each source and target language pair
machine translation tools designed to facilitate Hindi to [4].
English conversions. The importance of this theme lies in its produced a multimodal machine translation for the Hindi
potential to empower individuals and businesses operating in Visual Genome dataset between English and Hindi.
different locales. Whether it's cross-cultural communication, According to the author, a machine translation system
content localization, or accessing global information, performs better when it has a variety of input modalities
accurate and efficient Hindi to English machine translation is [7].
essential. Our exploration begins with a review of existing Direct machine translation is a straightforward translation
machine translation tools, examining their strengths and method that translates individual words from an input
limitations, and assessing their suitability for real-world source sentence into their corresponding words in the
applications. Highlight opportunities and improve the quality destination sentence after certain syntactical rearrangements
of Hindi-English translations. Machine translation has to maintain the language's sentence structure [16][17][18].
evolved significantly and Vanguard Machine Translation
(NMT) is at the forefront of this transformation. NMT 3. PROPOSED METHODOLOGY
models have demonstrated a remarkable ability to capture Machine translation from Hindi to English using Neural
linguistic nuance and context, revolutionizing translation Machine Translation( NMT) ways is a state- of- the- art
accuracy. in this article, we will examine how NMT approach that leverages deep literacy neural networks to
technology has been used to improve the quality of achieve accurate and contextually meaningful translation.
translations for Hindi speakers.

68
 Mallamma V Reddy et al / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 14 (5) , 2023, 68-70

NMT techniques are applied in this process

• Data Collection : Begin by collecting a substantial
bilingual dataset containing aligned Hindi and English
rulings. This dataset serves as the training material for
the NMT model.
• Data Preprocessing: Clean and preprocess the dataset by
removing noise, special characters, and formatting
inconsistencies. Tokenize the text into words or
subword units to handle verbal variations effectively.
• NMT Model Selection: Choose an NMT armature,
similar as the motor model, which has proven largely
effective in restatement tasks. Mills are designed to
capture long- range dependences in language and
exceed at handling environment.
• Training: Train the named NMT model on the
preprocessed bilingual dataset. During training, the
model learns to collude Hindi rulings to their
corresponding English restatements by conforming its Figure1: General architecture of converting Hindi text
parameters grounded on the input- affair dyads. to English text
• Attention Mechanism: NMT models frequently employ
attention mechanisms that allow the model to The above architecture diagram, each module is explained
concentrate on specific corridor of the source judgment in details
while generating the restatement. This attention medium • User Interface: The front-end server hosts the user
helps the model prisoner environment and ameliorate interface and handles user interaction
restatement quality. • Input Text: it manages the user input, ensuring it is
• Hyperparameter Tuning: Fine- tune the model's properly received by the system.
hyperparameters, including literacy rates, batch sizes, • Preprocessing: Responsible for cleaning and
and the model's armature parameters, to optimize normalizing the raw Hindi text to make it suitable for
restatement performance.
further processing.
• Evaluation Metrics: Use evaluation criteria like BLEU(
• Text Recognition: Recognizes Hindi text using optical
Bilingual Evaluation Understudy) to quantitatively
character recognition (OCR), then translates it into
assess the quality of restatements. BLEU compares the
English.
machine- generated restatements to mortal references.
• confirmation and Testing: Split the dataset into training, • Post Processing: Enhances the recognized text, which
confirmation, and testing sets. The confirmation set may include improving formatting, correcting errors, or
helps cover the model's progress during training, while refining the translation.
the testing set evaluates its conception capability. • Output Text: Generates the final translated output text in
• Deployment: Integrate the trained NMT model into English
stoner-friendly operations or interfaces, allowing
druggies to input Hindi text and admit English 4. RESULT
restatements accessibly. Input text: The input for a machine translation system from
Hindi to English consists of text or sentences written in the
The proposed using NMT techniques, machine translation input Hindi Text (figure 2).
from Hindi to English achieves impressive translation
quality, making it a valuable tool for bridging language
barriers in various domains,

3.1 System architecture

This architecture shows Hindi text is entered by users using
the user interface in this architecture. The preprocessing
service receives the input text after which it cleans and
normalizes it. The text recognition component recognizes
the prepared Hindi text and then translates it into English. A
post-processing service then polishes the translated content.
The system ultimately produces the better final translated Figure 2: Input Hindi Text
output in English. as shows in below figure1.

69
 Mallamma V Reddy et al / (IJCSIT) International Journal of Computer Science and Information Technologies, Vol. 14 (5) , 2023, 68-70

Output text: The translated text in English is what the 4. Deepak Khemani, “A First Course in Artificial Intelligence”, 2013,
pages 702-703
machine translation system produces Based on the input
5. M. D. Okpor, “Machine Translation Approaches: Issues and
given in figure 3. Challenges”, IJCSI International Journal of Computer Science Issues,
Vol. 11, Issue 5, No 2, September 2014
6. Wang, D., & Xiong, D. (2021). “Efficient Object-Level Visual
Context Modeling for Multimodal Machine Translation: Masking
Irrelevant Objects Helps Grounding.” In Proceedings of the AAAI
Conference on Artificial Intelligence (Vol. 35, No. 4, pp. 2720-2728).
Wagadiya N, Ravarta P English-Hindi Translation system with Scarce
resources. International journal of innovative research and
development.
7. ManishaRoy,MadhuraWalivadekar,PoojaKadam “Sentence
Validation Using Natural Language Processing”. Proceedings of IRF
International Conference, 23rd February 2014, Pune, India. ISBN:
978-93-82702-61-0
8. Shibli Syeed Ashrafi , Md. Humayun Kabir , Md. Musfique Anwar,
A.K.M .Noman, Gulshan, “English to Bangla Machine Translation
Figure 3: Output English Text System Using Context - Free Grammars” ,IJCSI International Journal
of Computer Science Issues, Vol. 10, Issue 3, No 2, May 2013.
5. CONCLUSIONS AND FUTURE WORK 9. Vishal Goyal, Gurpreet Singh Lehal “Hindi To Punjabi Machine
Translation System”. Proceedings of the ACLHLT 2011 System
The Hindi to English machine translation project has been a Demonstrations, pages 1- 6,Portland,Oregon USA.21 June
worthwhile attempt that shows how technology may help 2011.Association for Computational Linguistics.
people communicate in different languages. This tool has a 10. Ankita Agarwal, Pramila, Shashi Pal Singh, Ajai Kumar, Hemant
Darbari “Morphological Analyzer for Hindi – A Rule Based
lot of potential for promoting successful communication
Implementation” ,International Journal of Advanced Computer
and cross-cultural interchange between Hindi and English Research (ISSN (print): 2249-7277 ISSN (online): 2277-7970)
speakers in a variety of contexts, even though more Volume-4 Number-1 Issue-14 March-2014.
improvements are needed to improve accuracy. Regarding 11. David Peter S, Latha R. Nair” Machine Translation System for Indian
its potential influence and contribution to promoting global Languages”,International Journal of Computer Applications (0975 –
8887) Volume 39– No.1, February 2012.
connectivity, we are enthusiastic 12. Sudip Naskar, Sivaji Bandyopadhyay “Use of Machine Translation in
The Project future work will concentrate on advancing India:Current Status”, Jadavpur University,Kolkata.
translation accuracy, enhancing user experience, and 13. Holger Schwenk, “Continuous space translation models for
resolving potential prejudice and ethical issues. Our Hindi to phrasebased statistical machine translation,” In Proceedings of
COLING 2012: Posters. The Coling 2012 Organizing Committee,
English machine translation technology will constantly be Mumbai, India, pp. 1071–1080 (2012).
enhanced and expanded in order to better serve users and 14. Sainik Kumar Mahata, Dipankar Das, and Sivaji Bandyopadhyay,
encourage successful inter language communication. “Mtil2017: Machine translation using recurrent neural network on
statistical machine translation,” Journal of Intelligent Systems pp. 1–7
(2018).
REFERENCES 15. J. Nair, Amrutha Krishnan K, Deetha R, “An Efficient English to
1. Shachi Dave, Jignashu Parikh, and Pushpak Bhattacharyya, Hindi Machine Translation System Using Hybrid Mechanism”, 2016
“Interlinguabased English Hindi Machine Translation and Language
Intl. Conference on Advances in Computing, Communications and
Divergence,” Machine Translation, vol. 16(4), pp.251–304 (2001).
Informatics (ICACCI). IEEE. 2016, Sept, 21-24.
2. V Goyal, G S Lehal. “Advances in Machine Translation Systems”. 16. N. Wagadiya, P. Ravarta, “English-Hindi Translation system with
Language In India, Vol. 9, No. 11, 2009, pp. 138-150.
Scarce resources,. International journal of innovative research and
3. Anoop Kunchukuttan, Pratik Mehta, and Pushpak Bhattacharyya,
development.
“The IIT bombay english-hindi parallel corpus,” In proceedings of the 17. L. R. Nair, David Peter S., “Machine Translation Systems for Indian
eleventh international conference on language resources and
Languages”, International Journal of Computer Applications (0975 –
evaluation (LREC 2018), May 7-12, 2018, Miyazaki, Japan, isbn:
8887) , Volume 39– No.1, February 2012.
979-10-95546-00-9, (2018).

70