Regional languages give Indian startups a voice of their own

New Delhi: Kavita and Sirish Reddi who moved from the UK to Hyderabad in 2009 often wondered why voice-enabled virtual assistants from both Amazon.com Inc. and Google (Alphabet Inc.) focus primarily on the top 100 million English speakers. “There are not many local language equivalents of Alexa. Even Google has a few languages, but only for phone functions and search, not transactions,” Sirish Reddi points out. Sensing a business opportunity, the couple plans to launch Voxta, which they term as a “Mobile Alexa” voice assistant app in Indian languages.

“Voxta will be launched in Hindi, Telugu and Indian English and other local languages will be added gradually,” says Kavita Reddi, who along with husband Sirish is co-founder and director of the startup. Machine Learning—an Artificial Intelligence, or AI, technology—is at the core of Voxta’s products, according to Sirish Reddi.

“We use open-source Machine Learning frameworks with custom configurations tailored to our data and each language. Our voice bot framework is easily configurable to support natural language understanding for multiple domains and languages, and iteratively improves with usage,” he explains.

Voxta, though, is not the only Indian startup focusing on local languages. “Vernacular digital content is a big inflection point in India, and voice engines are the next big thing,” notes Jayanth Kolla, founder and partner, Convergence Catalyst—a digital technologies research and advisory firm.

Bengaluru-based Reverie and Liv.AI, for instance, are two such startups that also focus on delivering regional-language solutions through voice. Founded in 2010, Reverie works with mobile manufacturers, governments, online retail, e-commerce marketplaces, travel, media and entertainment companies, banks and financial services, “and the developer community”, according to co-founder and chief executive officer Arvind Pani.

Reverie’s Language-as-a-Service (LaaS) platform, he explains, comprises automated content conversion, cross-language content search and discovery, text and voice input interfaces and language analytics, “delivered through a set of APIs (application programming interfaces)”.

Pani, though, believes that voice and text complement each other. “The intent of a user query is derived form text. So it is the marriage of voice and text that enhances the overall user experience,” he explains.

Liv.ai is another AI startup with a platform to convert speech to text in 10 Indian languages—Hindi, Bengali, Punjabi, Marathi, Gujarati, Kannada, Tamil, Telugu and Malayalam.

The business opportunity is big. Indian language internet users will drive the next phase of internet adoption in India and will be more than 2.5 times of English internet user base by 2021, according to analysts.

Further, last year, Google said voice search accounted for a staggering 30% of total search on Google in India and search in Hindi grew at a whopping 400%. Moreover, 150 million Indians will use a voice assistant by 2020, and 50% of all search will be on voice, according to ComScore.

Indian language internet users are expected to account for nearly 75% of India’s internet base by 2021, according to a 2017 Google-KPMG report. This only spells good news for these local language startups.

Kolla believes Indian players in this space are superior to companies like Google and Amazon even though the latter have scale and money to invest. He concludes, “Local players have better focus and for them this is their entire business. Entire teams are aligned for this and they are thinking of domestic solutions.”