India’s automotive industry is going through a huge revival. As the automotive industry witnesses a remarkable surge in passenger vehicle sales, an Indian firm, Reverie Language Technologies, is looking at taking center stage with its innovative approach to building multilingual speech interfaces for in-car voice assistants.
The surge in passenger vehicle sales, reported by the Society of Indian Automobile Manufacturers (SIAM) with a staggering 26.7% increase in fiscal year 2022-23, has spurred a parallel growth in the adoption of advanced in-car voice assistants.
Says Vivekananda Pani, Co-Founder- CTO, Reverie Language Technologies, “The remarkable surge in sales is complemented by a rising trend in the adoption of voice assistants in automobiles, driven by the expanding capabilities of in-car voice systems. These voice assistants enable drivers to effortlessly perform a wide range of hands-free tasks, including navigation, calls, information retrieval, ordering, and control of various car systems.”
Historically, the in-car infotainment system has traversed through the analog era (tape cassette players, basic dashboards), Digital Emergence (GPS Navigation, rare view mirrors in high end models). However, just as India has leapfrogged in other areas in terms of technology adoption, we can expect the same in this sector too. “As the economy and mobility accelerated post the pandemic lockdown and larger dependency on virtual assistants for navigation, information and entertainment we foresee the Smart Era paving the way for connected vehicles and virtual assistants being modified and adopted for all categories of vehicles including two wheelers,” states Vivekananda Pani.
Reverie had adapted to the changing demands of the Indian market by focusing on the aspects that are unique to India. For example, recognizing the diverse and noisy environments within vehicles, Reverie focused on the need to consider variations in utterances, accents, and dialects. External factors like traffic noise and weather changes, coupled with internal elements such as passenger conversations and in-vehicle systems, pose formidable challenges. Reverie’s approach involves training speech engines to navigate these interruptions effectively.
Addressing the constraints of limited memory resources in automotive systems, Reverie’s multilingual speech-to-text model emerges as a technological marvel. Vivekanand Pani details the four components powering Reverie’s in-car voice assistant – the wake word, speech to text, natural language understanding, and text to speech. The delicate balance between low memory consumption, CPU optimization, and accuracy ensures a seamless and memory-efficient user experience.
In the future, voice assistants will become an invaluable part of every vehicle, in enhancing both convenience and safety on the road. As virtual voice assistants become an integral part of the driving experience, automobile manufacturers will be quick to adopt these innovations to differentiate their brands. The use of voice commands allows users to interact effortlessly with their vehicles, ensuring a safer and more connected driving experience.
Reverie’s solution for MG Comet exemplifies the practical implementation of multilingual interfaces for offline voice assistants. From controlling in-car systems to tuning radios and enabling hands-free calls, the voice assistant understands a mix of Hindi and English commands, catering to diverse user preferences. The key use cases span from music and radio control to system commands and specific vehicle controls, showcasing the versatility of Reverie’s multilingual interface.
In a multi-lingual market such as India, Reverie Language Technologies has a substantial market opportunity driven by the rise in voice technology adoption, with a specific focus on multilingual solutions and in-car voice assistants. The company’s potential to cater to linguistic diversity and tap into the growing trends in the automotive industry, smart devices, and IoT positions it well for future growth.