(Chinese Version)

On October 4th, 2011, Apple introduced Siri on the iPhone 4S, becoming the first smartphone maker to integrate voice recognition function in smartphones. In the following year, Siri began to support Mandarin and Cantonese. Apple’s action led to a boom of voice smartphone recognition applications around the world.

Five years after the launch of Siri, various voice-control scenarios have already been developed. As a matter of fact, human-computer interaction based on voice recognition technology, has already become an essential function in all smart home devices.

Although both Chinese and overseas tech giants have been investing heavily in this emerging new business, few users seemed to trust their voice assistant and use voice recognition function as frequently as touchscreen control.

The current development of voice recognition technology in China and abroad

Three players are dominating the global voice recognition sector, namely, Apple (Siri) and Samsung (S-Voice), Google (Google Now) and Microsoft (Cortana).

In China, IFLYTEK is undeniably the leading voice recognition service provider. JD Domestic appliance manufacturers such as SmartCloud, Smartisan, Changhong, Skyworth have all been cooperating with IFLYTEL on voice recognition technology.

While later-comers such as Tencent and Baidu caught up through abundant capital and rich resources, young service providers such as AI Speech, SinoVoice, Unicound attracted wide media coverage due to the high accuracy of their services. At the same time, other small-size teams and companies were also developing pretty fast.

Voice recognition in smartphones

At first, Siri could only read users’ text massages, recommend restaurants, forecast weather, awake users at set time, arrange schedules, search information, etc. Later on, as technology progressed, Sir were able to learn new accents and even converse with users.

At the same time, lots of app developers have been introducing voice control to their apps. These apps include WeChat, Weibo, 365 Calendar, iReader, 91 Panda Reading, etc., to name just a few.

Voice recognition in smart home devices

With the rise of IoT industry, lots of tech companies have been cashing in to make a fortune.

In overseas market, Apple, Google and Microsoft have all been eyeing on the sector:

On the WWDC held this June, Apple unveiled its own smart home platform called Homekit. With the help of Siri, users will be able to remotely control smart home devices.

Recently, Microsoft announced that it would integrate Microsoft's intelligent assistant, Cortana, into Insteon’s Windows Phone app.

Google started early and entered the industry by acquiring Nest Labs, a manufacturer of smart thermostats and smoke alarms, last year. On the I/O Developer Conference held this May, Google unveiled its new IoT platform Brillo and further upgraded Google Now.

In China, it was quite common to enter the industry by either providing smart home devices based on voice interaction technology or developing smart home robots based on human-machine interaction technology.

Is voice recognition really useful enough?

Although voice recognition function has already become an essential function of many apps or smart devices, the user frequency remains relatively low. The inconvenient fact is that most users open their voice recognition software merely for fun.

How come? The following four factors might help explain:

1. The accuracy remains low
Although Apple, Google and Microsoft have been constantly upgrading their voice recognition software, there is still a long way to go before the accuracy of voice recognition reaches 99% or 100%. Such factors as users’ mood, physical condition, noise, context are all influencing the accuracy of any voice recognition software.

2. The complicated Chinese language
Chinese is too much more complicated than English. Even the same sentence spoken with different intonation, pronunciation and accents might have completely different meanings, let alone the large number of argots and metaphors in Chinese. No wonder overseas developers find it extremely difficult to carry out a voice recognition system that can support Chinese and accurately understand and respond to Chinese. In this point, Chinese developers ought to take part in and endeavor to develop a smart enough Chinese voice recognition system.

3. Limited number of high user-frequency scenarios
Although voice recognition technology has already been pre-installed in various kinds of devices, such as cars, smart home devices, smartphones, etc., the number of high user-frequency scenarios of voice recognition technology remains limited. As a matter of fact, voice recognition function is to a large degree like a “chicken rib”-tasteless when eaten but a pity to throw away. In other words, the user experience of voice recognition function in most devices is far from satisfactory at present due to the different accuracy, reaction time and power consumption rates, but it is a pity to buy a device equipped with voice recognition technology without using it quite often.

4. More than a substitute
If voice recognition technology is simply developed as a substitute for touchscreen control and buttons, then consumers will never find it attractive to turn to voice control. Real innovation needs to be made to figure out a really user-friendly voice recognition system.

Widespread as voice recognition technology is, the user frequency remains relatively low. It is fair to say the technology hasn’t been really that “widespread” yet.

Yet, nobody can deny that voice recognition technology has already become the must-trend of human-machine interaction industry, and is going to gradually infiltrate into people's daily lives in the near future. The smart hardware industry might even be transformed by the technology.

Whoever first carry out the most user-friendly voice recognition system and achieve real breakthrough in terms of accuracy, reaction time and processing performance will certainly make a fortune in this booming industry.

[The article is published and edited with authorization from the author @zhidxcom, please note source and hyperlink when reproduce.]

Translated by Levin Feng (Senior Translator at ECHO), working for TMTpost.