CIOInsider India Magazine

At times using the voice search feature from Google happens to misinterpret what the user is actually pronouncing. This is quite common for the people belonging to diverse ethnicities, as their slang could make the word that they intended to pronounce end up appearing as an entirely different word with different meaning. That’s why Google, which is ardently working on AI, is inventing tools and features that allow users to use the voice search, including its other features, to accurately detect and serve people’s requirements in the language of their preference.

At this year’s 8th Google for India event, Google unveiled some of its most cutting-edge tools and apps aimed at creating AI-powered and India-first solutions. On the other hand, tech giant revealed its plans for launching courses on YouTube in India.

Google suggest With AI - Searching with Senses
With ‘Google suggest With AI’, users can create search experiences that correspond to the ways they perceive the world through sight, sound, and language. In India, people are increasingly adopting voice search to find information online and their phone cameras to take pictures of their surroundings.

Multisearch - Explore Information Visually
Google’s Multisearch combines words and visuals to make visual search seem more natural. All users can do is by merely using the screenshot feature or the camera in the Google app to capture an image and thereafter add words to their query. Users can use this tool to search for a specific item using a variety of criteria and to combine text and images in their queries. For online shoppers who might need to locate things but only have a photo of the item and no description, this is especially helpful. Or it can direct you to a location where a specific meal might be served. In India, Multisearch is available in English; effective in the next year, it will also be available in Hindi and other Indian languages.

Bilingual Search Result Pages
To make searching and exploring content easier for people who speak many languages, Google is creating bilingual search results pages. It will provide users with high-quality and pertinent content in their native language along with English results using cross-language search technology and machine learning-based translation models. Google claims that this innovation is a first for India. Next year, this feature will also be available in Tamil, Telugu, Marathi, and Bengali in addition to Hindi.

Speech Recognition Software for Hinglish Speakers
Since India embraces a diversity of languages, Google has got that covered as well. By enabling people to search for information and do tasks online by speaking in their own language, voice-based technologies have revolutionized India. According to the company, twice as many Indians use voice search on a daily basis as the average person worldwide.

With that said, Google unveiled A new voice recognition software that is more adept at understanding Hinglish. How the software achieves this is by deploying a cutting-edge speech recognition system inspired by neural networks which is capable of understanding a speaker's accent, background noise, context, and speaking style.

Project Relate: Recognizing Irregular Speech Patterns
Project Relate, an app adapted to irregular speech patterns, is the product of years of research. Currently, Google is piloting the app in India with English users, and is looking towards adding Hindi users on the platform by 2023. Through the research, Google hopes to increase the usability of speech-recognition tools like Google Assistant. Project Relate is taught to recognize the distinctive speech patterns of persons with non-standard speech, as a result of years of research. In India, the testing with the app has only been with English speaking users thus far and the tech giant is planning to conduct the same with Hindi speaking folk as well. With this initiative, Google hopes to increase accessibility for all users of products that use speech recognition technology, such as Google Assistant.

Digilocker - Easy Access of Digital Documents
Google announced its partnership with the National eGovernment Division (NeGD) to give consumers easy access to their legal digital documents. The Android Files by Google app will provide users fast access to this. The foundation of this service is security and privacy. Files by Google docs are kept in a secure environment on your smartphone and can only be accessed with a special lock screen verification.

A specific AI model is used by Google to find and identify important documents on the user’s phone. Following that, OCR technology and machine learning models are used to recognize the pertinent text and combine this information into an easy-to-read view. These models are all offline and accessible only through a specific device.

Detecting Fraudulent Activities on Google Pay
It's difficult to overstate the impact of UPI and digital payments on the millions of individuals who routinely engage online as Google Pay celebrates its fifth birthday in India. The Google Pay safety engine employs novel techniques to examine the transactional habits of millions of users as a billion-node graph.

This technology assists in the detection of fraudulent behavior, and then notifies the user in their favorite language using a sophisticated language engine. Google stated that the app "will occasionally vibrate to capture your attention."

Additionally, Google Pay is gaining a useful new feature that, according to Google, makes retrieving your transaction history easier than ever. You can now use natural language to ask Google Pay questions, such as "Show me how much I spent on coffee last week."

AI for Health
To help one lakh Indian mothers, Google is now deploying AI. In India, it might be difficult to read handwritten prescriptions, hence, Google is using AI to help. However, perfecting this approach still remains underway.

Apollo is collaborating with Google on a project to aid our artificial intelligence in x-ray tuberculosis detection. AAI is also being used by Google to create a clinical intelligence engine that can be used to identify prevalent disorders.

AI for Farmers and Agriculture
Google is employing remote sensing technology to provide an overview of India's agricultural potential. They have also created a landscape comprehension model, which was created in Bangalore. In Telangana, it is undergoing testing. It can recognize both natural and artificial water bodies, as well as fields. The program will also be able to identify events that happened in the fields, such when the crop was seeded.

Project Vaani
Google intends to work with the Indian Institute of Science to develop an AI/ML model. Language samples from 773 Indian districts were to be gathered for the Project Vaani initiative. The data gathered and generated as a result of Project Vaani will be open-source.