Why type to search for nearby restaurants, hotels or even toilets using a tiny keyboard when you can just speak to your phone. That’s the premise of the new, free Chumenwenwen voice search application that is available on multiple platforms including WeChat, the most popular mobile messaging app in China.
“We now have more than 200,000 users, and the figure is rapidly expanding now “said Yizhou Zhu, 26, the co-founder of Shanghai-based startup Mobvoi. Graduated from Stanford University, he chose to devote himself in startups focusing on mobile voice search. The product Chumenwenwen means “go out and ask” in Chinese, which indicates its two salient features: mobile and voice search. The motivation to launch a start-up and cash in on the growing need in mobile voice search in the world’s largest Internet market was not uncommon.
The Office of Mobvoi
Voice search is becoming more popular then ever. This is no surprise as smartphone penetration and usage continue to grow. According to a recent research done by the Telemetrics. 50 percent of the users in the US start to use a mobile device to search. For certain type of searches such as the nearby restaurants, such figures are even higher. The famous restaurant ratings& reviews service Yelp reported that 59 percent of all their searches are on mobile in Q2 2013.
“But that’s not the only factor behind the uptick in mobile voice search,” said Zhu. He cites the demand for better user interfaces, especially from those who don’t want to rely solely on a touchscreen to interact with their phone. In addition, the increasing prevalence of laws restricting the use of cell phones while driving worldwide, especially US, has fueled demand for these voice-based apps. Bluetooth headsets allow drivers to talk hands-free, but many people want access to e-mail and text messages while driving, too, he notes.
Technology giants Apple and Google have already detected this trend for long. Apple acquired Siri and added it to iPhone as a reaction. They understood the future of search on mobile would increasingly or even overwhelmingly be voice search. Google also acknowledges that people want more easy-using interfaces. Speech input has already been set as the primary interface on Google Now. However, to either company, Chinese language is a great obstacle barring on the beginning of their long journey.
“In China, users have had a hard time using Siri to get precise answers to certain questions. Siri often fails to deliver a satisfactory answer.” said Chaofei Fan, 25, now a software engineer at Chumenwenwen. “In this case, the mobile voice search app Chumenwenwen wants to be an antidote. To get Chinese voice recognition work, we spent a lot of time letting the machine learn about this fascinating language — its tonal characteristics, the different accents, pinyin representations of Chinese characters…the list goes on and on. Chinese language tone is unique. The subtle difference in tones can mean completely different things, for example”Can Guan” in Chinese can mean restaurant with the second character in its third tone whereas it means visit when the second character share the first tone. Therefore, we have our unique innovative system that can understand its meaning according different contexts.”
Besides the difficulty in recognizing Chinese language, it has to work out other difficulties faced by all the mobile voice search applications. “Voice search queries are more complex than traditional text searches, because it involves more natural language understanding. ” Fan added. “Chumenwenwen answers your questions in three steps: first it does speech recognition to convert your voice into text; the text is then passed to semantic analyzer to understand your intention; finally with your intention Chumenwenwen searches relevant domains to find the most accurate answer for you. Among these three steps, Chumenwenwen specializes in natural language understanding. We are seeking to provide more accurate search results by both understanding what users say and refining the results based on the context such as location and other data we knew about users.”
For example, when you are asking for air ticket information on Chumenwenwen, you can just speak as if you were talking to a travel agency and Chumenwenwen will get all the details including time, destination, price, airlines, etc. You can talk in any way you like: such oral expression as Di Du or Mo Du can be instantly understood as Beijing or Shanghai.
Mobile will bring a new uprise of voice search. With many users making restaurant reservation or purchasing deals on mobile devices, voice search seems like a natural progression. “We believe voice search is going to aggressively expand into every function of human lives. One of the key issue is to let mobile phone understand more natural language and complex queries. And that’s precisely what we are striving for” Zhu affirmed.
Leave a comment