Bhopal: During the APAC National Cloud & Cyber Security Summit held in Bhopal, Amitabh Nag, the Chief Executive Officer (CEO) of Digital India Bhashini Division (Bhashini), Digital India Corporation, shed light on how the AI-powered language translation tool, Bhashini, is contributing to collaborative engagement, research, and the empowerment of end-users, by enabling real-time translation of Indian languages.
“The mission’s objective was to transcend the language barrier. The target was to look at the 22 languages which were part of the Constitution,” Nag said.
Bhashini’s Vision
- Fostering collaborations and partnership programs with both national and international organisations.
- Driving collaborative research, enhancing commercialisation awareness, and supporting capacity building.
- Developing a long-term strategy for creating sustainable Indian language technology solutions and fostering a robust technology ecosystem.
- Promoting the adoption of Indian language technology solutions to facilitate easier access to the internet.
- Leveraging economies of scale in Indian language technology solutions, applications, data sets, and AI models.
- Supporting cutting-edge research in Indian language technologies, encompassing both evolutionary and revolutionary advancements.
- Promoting and facilitating the generation of Intellectual Property (IP).
Areas of Focus: Unraveling the Magic in Speech Tech, NMT, and OCR
- Automatic speech recognition: It is a technology that transforms spoken language into written text, enabling voice-controlled systems and transcription services. Bhashini can currently translate 12 languages with expertise, along with the English language.
- Speech technology: This includes innovations ranging from Automatic Speech Recognition (ASR) for transcribing speech into Text-to-Speech (TTS) with the aim of converting the text into a spoken language that would seem lifelike.
- Neural Machine Translation (NMT): It represents a cutting-edge approach to automated language translation, harnessing the power of artificial neural networks. This method yields translations that are not only more fluent but also more accurate when compared to conventional approaches.
- Optical Character Recognition: It is a technology designed to convert printed or handwritten text, as well as images, into machine-readable text.
Transforming translation dynamics: DIBD’s integrated approach
- Framework centralisation: Digital India Bhashini Division (DIBD) has established a facility that centralises translation AI models and associated research work. This enhances community access to the data, facilitating its utilisation for further implementation.
- People-powered model training: Through the Bhashadaan movement, DIBD facilitates the training of models via widespread public participation on a large scale. This empowers citizens to actively engage in the mission of creating the world’s largest AI use case model.
- 70 research institutions aligned: DIBD enables research and partner institutions to leverage exercises and develop innovative solutions through startups/projects, aiming to simplify translation for stakeholders.
- Empowering end users: Following an intensive process of training models and implementing technical interfacing, Bhashini’s APIs are made available to end users, enabling them to commence their work on translation segments through Text-to-Speech (TTS), Automatic Speech Recognition (ASR), and various other translation types.
Breaking Language Barriers and Fostering Multifaceted Engagement: Bhashini’s Ecosystem
Bhashini’s ecosystem plays a pivotal role in realising the vision set forth by MeitY, effectively breaking down language barriers. This collaborative framework extends its reach to central and state governments, aligning with language-specific initiatives to fulfil the overarching mission.
In partnership with Language Missions, Bhashini actively engages in the identification of data sources, data collection, content creation, and the strategic execution of crowdsourcing initiatives.
Further, academia and research groups find a valuable space within this ecosystem, participating in research and development activities focused on advancing language technologies.
Startups contribute by developing innovative multilingual applications and services, while the industry at large plays a role in the development of open-source software, offering storage solutions, and contributing computational resources for model training.
Data collection and curation companies are integral contributors, responsible for the collection, validation, and curation of datasets. Publishers are enlisted to provide essential data sources for building datasets and models.
Lastly, individuals are invited to actively contribute to the mission through participation in crowdsourcing initiatives, thus creating a comprehensive and collaborative environment.
Bhashini for Agriculture: Driving Next-Generation Technology
Bhashini is extensively utilised by central and state-level agriculture departments to implement automated and technology-driven methods. Serving as a crucial framework, Bhashini facilitates a range of applications, from chatboxes to financial transactions, addressing the essential needs of farmers in their last mile.
- Through Chatbox: Through Bhashini-empowered chatboxes, farmers gain access to accurate information and beneficial schemes.
- Land Records and Translation: With support for vernacular languages, Bhashini ensures that land records are readily available, providing farmers with improved financial options.
- Other Farmer-Centric Operations: The availability of information in regional languages significantly impacts the daily lives and operational needs of farmers.
Whether breaking language barriers or driving next-generation technology in agriculture, Bhashini exemplifies the transformative power of language technology, getting one step closer to a more connected and accessible world.
