Case study for in-car system

The Challenge

The company needed to develop more than 20 types of multilingual speech recognition data based on the industry's growth. With the pressure on today's automakers in the industry content increasing year by year, the in-cabin voice recognition system remains a major complaint of vehicle owners. A very important issue that almost all global automakers face is the implementation of localization of in-vehicle systems. Due to the complexity of multiple languages, it is difficult for car company teams to handle and manage the large amount of linguistic data. Not only do they have to ensure voice collection in various driving environments, but they also have to clearly capture all aspects of language content including weather conditions, road types, and many other driving scenarios. Since the car company's engineer team does not have the corresponding language background, the collection of speech data has to be handled by a technology company with deep language expertise.

Solution

As an automotive solution provider with several years of experience, we offer a full range of services for localized data collection, testing, and validation. Our professional team can work directly with the engineering teams of automotive companies to implement and optimize the execution of solutions to meet the high quality training data for all target languages. At the same time, our rich team resources can recruit a large number of local native speakers to record various environmental languages to achieve guaranteed output of data quality and quantity.

Our professional TTS service capability can provide professional evaluation services during data collection to help improve the accuracy and naturalness of speech in driver command scenarios.

Meanwhile, through our many years of industry experience in driving, we have a professional language team as well as an AI intelligence center that can provide various professional guidance services for our processes during implementation to ensure language expertise, localization and other multifaceted capabilities. And keep the consistency of various language standards to balance the output results of all training data. Help all the team members to support a large amount of localization work content.

Results

Through our collaboration, the team from the car manufacturer was able to work very efficiently with our team with many years of experience, focusing on improving all core stages of system development, extensive data collection capabilities and high quality data annotation capabilities, effectively helping each other to solve their needs and avoid possible problems. With our solution, the company has completed advance planning to move into a more comprehensive multilingual speech recognition system development, effectively expanding into new market opportunities.