In February 2013, Ajit gave a TED talk on a new linguistic structure called FreeSpeech, and an algorithm called the FreeSpeech engine. FreeSpeech was created in response to a need felt among children with complex communication needs to communicate in multiple languages. FreeSpeech is a semantic map of pictures that represent words. These words are linked together in pairs of question/answer relationships, and have pictorial markers for tense, number etc. applied on top of them. The structure's stated aim was to capture 'meaning' rather than 'surface form' of language, and then use a set of algorithms to generate grammatical English out of it. FreeSpeech is language-independent, meaning it uses pictures of objects, actions, and even abstract ideas like the past tense, instead of English words like "I" or "eat". Some shown on the TED talk include a bowl of soup, a person talking, and a small clock with an arrow pointing backward to signify the past tense of a verb. FreeSpeech was tested widely among children with special needs such as autism and language impairments, to teach them grammar. It also saw use by children in the Deaf community to learn literacy—a task complicated by the unique grammatical structure of American Sign Language, which is very different from English. Beyond applications in disability, the Indian government was in talks to license this program to use in some of its schools to teach English, through the Aakash tablet project.
In 2011, Ajit was named the MIT TR35 "Innovator of the Year" for India. His citation for the award mentioned his work on low-cost speech generating devices:
There are an estimated 10 million people in India who suffer from speech impediments. They may not have speech but they have a lot to say. And they can benefit from Narayanan's device. Avaz is a communication device for people with speech disorders such as cerebral palsy, autism, mental retardation, and aphasia. It works by converting limited muscle movements, such as head or finger movements, into speech. His invention broadly falls under the category of Augmentative and Alternative Communication technologies. Though speech generating devices are effective, most AAC devices aren't within the reach of the speech-impaired persons in the developing world and they mostly generate speech in English. This is another deterrent which has prevented these devices from becoming as popular in the developing world. Narayanan's innovation lies in bringing down the cost of the device to one-tenth of the price of similar devices, making it affordable to a wide swath of the Indian disabled population, and making it available in Indian languages.
Narayanan's Invention Labs, based in Chennai, designed Avaz to be not only cheap but also capable of supporting multiple languages. "The average young person in India speaks and uses three different languages every day," Narayanan points out. By working directly with Asian hardware manufacturers, he has been able to bring the cost of an Avaz down to around $800, compared with $5,000 to $10,000 for a single-language device in the United States.
