Application under development to help people know about agriculture in Bengal dialects
Two institutes in the country have teamed up to develop a speech recognition application that will help less educated people in West Bengal understand different aspects of agriculture and economy in their own dialects.
Maulana Abul Kalam Azad University of Technology (MAKAUT) said it is collaborating with Indian Institute of Science (IISc) Bangalore to release the app which will be used by people for free.
“Initially, about 12 dialects of North Bengal and South Bengal are considered for this research,” said MAKAUT Technology Cell Head Pritimoy Sanyal.
The entire research work is divided into two parts. One part is collecting words used on agriculture and economics in various dialects and the other part is making the computer understand them through artificial intelligence (AI), he said.
After the project is completed, people can ask questions about agriculture and economics in their dialect through the app and get answers in the same dialect.
“This is the first such work on Bengali dialects,” Sanyal said.
MAKAUT Vice-Chancellor Saikat Maitra said a committee of Bengali language experts has been formed to help scientists develop the app.
“Since we are trying to include almost all Bengali dialects in this app, we might also think of those used in Bangladesh in the future,” Mitra said.
While the majority of Indians depend on agriculture, a large portion of them are not educated enough to understand the industry and economic information available on the internet as it is mostly in English.
With the rise of digital services, the development of digital technologies such as voice recognition in the languages of the poor is urgently needed, the IISc said.
To solve this problem, IISc scientists started a project called RESPIN with nine languages and their dialects from all over India – Hindi, Bengali, Marathi, Telugu, Bhojpuri, Kannada, Magadhi, Chhattisgarhi and Maithili. MAKAUT works on the Bengali language.
The approach is to generate and collect domain-specific speech data for these nine Indian languages. This will enable tech companies to create domain- and language-specific voice interfaces that are affordable and sufficiently accurate, the Bangalore-based institute said on RESPIN. “Voice recognition in Indian languages has the potential to help illiterate Indians use digital financial services by simply asking out loud for a service they need, including money transfer, payments and reporting stolen ATMs. or lost,” he said. The project is funded by the Gates Foundation of America.
(This story has not been edited by the Devdiscourse team and is auto-generated from a syndicated feed.)