Swecha, a non-profit organisation engaged in the free software movement in the two Telugu States of Telangana and Andhra Pradesh, has developed a Telugu voice assistant to train computers and phones to speak like humans with different dialects.
Named ‘Swecha Gonthuka’ (Voice of Freedom), the developers are in the process of collecting Telugu voice samples to amass 10,000 hours of voice data by the end of this month.
As part of this unique initiative, the Artificial Intelligence for Telugu concert was held where singer and composer Ram Miriyala donated his voice sample. He said, “It was a special experience, the voice you all gave means a lot for AI.”
“An AI for Telugu concert is hosted to encourage public participation and capture the essence of local dialects. The idea was to ask people to participate, simply donate their voice, or even better, as a family, by recording 50 Telugu sentences. This will aid in building a dataset that will power the next generation of AI in Telugu. The results will be showcased at the Swecha AI Conference in March,” says Kiranchandra of the Swecha Foundation.
Swecha Telangana aims to provide global software solutions to locals with a free software development model. It has already created AI “Chandamama Kathalu” (Chandamama Stories) engaging over 8,000 participants.
The Telugu text-to-speech system integrated with the desktop will help people, who cannot read, to ‘listen’ to the text, just by pointing at it. Swecha is a free software project driven by volunteers. Most components of the projects are licensed under the GNU General Public License.
“Most of the data used by large companies isn’t available to the majority of people. We think that stifles innovation. So, we have launched a novel initiative called ‘Common Voice’, a project to help make voice recognition open and accessible to everyone,” a company spokesman said.
Follow us
