iFlytek's voice cloud passes ten million users

Background

An input method is the layer a user uses to enter text into other apps. On mobile devices, screen and keyboard space is limited, so speech can be an attractive addition. A voice cloud, in general, moves speech processing to remote servers. A device sends audio to the service, and the service returns text or a command. This lets a phone or computer offer speech recognition without carrying all the computation locally. iFlytek described its voice input method as the world's first intelligent voice input method implemented through cloud computing.^[1]

What happened

In December 2011, iFlytek's own news page said the total user count of its voice cloud had passed ten million. The wording was total users, a cumulative measure rather than a measure of daily or monthly activity. The announcement concerned the voice cloud, while the input method was the product iFlytek described as cloud-based; the two were reported together with the product's awards and official recognition.^[1]

How the product worked

The input method combined voice, handwriting and Pinyin input, and let users switch among them on the same interface. Users could dictate text messages, microblog posts, QQ chats and web addresses.^[1] Mixing input modes is practical because different tasks suit different methods: speech can be faster for a message, handwriting can help with characters that are hard to type, and Pinyin remains a familiar keyboard-based route. A cloud-based design also means the recognition work can sit on remote servers rather than entirely on the device, though it depends on a network connection. Dictating a web address shows that the product was meant for general text entry, not only for chat.

Awards and recognition

In April 2011, the input method won first place in the Developer Planet innovative application category at the Global Mobile Internet Conference. It also won an award in the 2011 Mobile Internet Application Innovation Competition.^[1] In September 2011, MIIT officially recognised the voice cloud as a cloud computing solution based on secure and controllable hardware and software.^[1]

What it meant

For iFlytek, the ten-million figure gave a concrete measure of adoption for a service it presented as cloud-based. The awards and the MIIT recognition supplied separate forms of validation: one from a conference and competition, the other from MIIT. Together they framed the input method as both a consumer product and a cloud computing solution built on secure and controllable hardware and software.^[1] The episode is a record of how an input method was being packaged as a cloud service, with the user count, product awards and official recognition reported as connected pieces of the same push. For ordinary users, the significance is that speech recognition was presented as part of everyday text entry rather than as a separate demonstration.