Flipkart starts voice search in Hindi and English to tap next 200 mn users

Launches Voice Search capability across the platform after the introduction of Voice Assistant in the grocery last year

E-commerce giant Flipkart has introduced ‘voice search’ on its platform to enable customers to search from a pool of over 150 million products across more than 80 categories through speech. Currently available in Hindi and English, Voice Search will enable faster onboarding of customers from smaller towns and simplify their e-commerce journey.

The gradual rollout of ‘voice search’ since January this year is witnessing impressive adoption across the country with initial trends showing over 5 million queries a day. The launch of  ‘voice search’ comes after the introduction of  ‘voice assistant’ for Flipkart’s grocery store last year. With the introduction of this capability, consumers can now discover and buy products using voice commands in multiple languages, starting with Hindi and English across the platform.

“Flipkart understands the heterogeneity of the Indian market and is committed to solving for the next 200 million users online while catering to the needs of all users to make e-commerce more inclusive,” said Jeyandran Venugopal, chief product and technology officer, Flipkart. “Voice search, along with some of the other initiatives by Flipkart, will make the digital commerce experience more convenient, accessible and seamless for the new wave of online shoppers.”

More than 75 per cent of internet users in India coming from non-English speaking backgrounds and a majority of them reside in non-urban and rural areas. Flipkart said it becomes imperative to solve the language barrier for enabling access to the benefits of the internet, including e-commerce. According to a joint study conducted by Bain & Company and Flipkart in 2020, online shoppers in tier-2 cities make up nearly half of all shoppers and contribute to three out of every five orders for leading e-retail platforms.

Flipkart has introduced Voice Search to enable customers to discover and buy products using voice commands in Hindi and English and a blend of both. The voice search is capable of helping users search through the product catalogue using colloquial commands thereby reducing the friction points in their online shopping experience.

Studies conducted by Flipkart to understand the new-to-internet consumers or Bharat indicate that they need assistance in online shopping and seek easy discovery of products. Also, searching through voice is both natural and quicker as it is 3x faster than typing in English and 5x faster in Hindi, as per independent studies.

Therefore, voice can play an important role in bringing clarity, confidence, and comfort for new-to-internet consumers. It also improves Flipkart’s ability to understand the needs of these new users effectively, who are different from the metro and Tier-1 users and make their e-commerce journey simpler. This also gives sellers a great opportunity to build relationships with these new sets of users and make in-roads into larger markets, thereby benefiting the entire retail ecosystem in India.

To enable voice search in Hindi and English, Flipkart deployed a host of technical capabilities. These included automatic speech recognition, natural language understanding, and text-to-speech for Indian languages, built by Flipkart’s in-house engineering and data sciences team. Once a user gives a command, the automatic speech recognition recognises the voice and converts it into text. Here, the natural language understanding tries to identify the intent and entity from the keywords given in the command. These keywords are then picked up by Flipkart’s search engine, which fetches results for the customers in a fraction of a second.

Flipkart’s voice search merges these technologies with its existing text-based search capability. This is trained on years of Flipkart data to offer a powerful voice search platform that can understand natural language-based long and complex queries made by users in different accents.