Tech giant Google has joined forces with Bengaluru’s Indian Institute of Science (IIS) for Project Vaani, an initiative aimed at open-sourcing research models and datasets to aid developers in creating solutions with India-specific speech data and local information.
Under this partnership, the initial speech data set will encompass 4,000 hours across 38 languages. Manish Gupta, head of Google Research India, highlighted that out of the 125 languages Google is committed to supporting, 75 lacked digital data accessible to AI researchers until now. Through Project Vaani, digital data will be made available for the first time in certain languages, leading to potential innovations in these previously data-scarce languages.
Ambarish Kenghe, vice president of Product at Google Pay, emphasized the significance of this large language model for the AI community in India. The PaLM API, MakerSuite, and features on Vertex AI will provide AI experts with access to the model, benefitting the more than 60 generative AI startups in the country.
Furthermore, Google is extending its support to developers working on delivery-related apps through Plus Codes, which offer location information openly. Additionally, Google plans to launch a Trusted Tester program for developers focusing on healthcare AI models to aid in the identification of medicine names.
This recent collaboration with IIS is part of Google’s ongoing efforts to cater to India-specific requirements. Google will also introduce an accelerator program for the Open Network for Digital Commerce (ONDC), assisting sellers on the platform in scaling their businesses. As part of this initiative, Google Cloud will roll out a startup credits program, enabling ONDC startups to apply for a $25,000 grant.
Following a meeting between Google CEO Sundar Pichai and Prime Minister Narendra Modi during Pichai’s recent US visit, Google announced its plans to establish a global fintech operation center at the GIFT City in Gujarat. Furthermore, Google’s Indian Languages Program was launched earlier this month, offering language support to small and medium-sized vernacular news publishers in English, Hindi, Kannada, Tamil, Telugu, Bengali, Malayalam, Gujarati, and Marathi.
