How to work with Siri: local voice recognition + cloud computing Services

Apple's voice assistant service Siri is stylish and smart and, in some cases, useful, the ZDNet website said today. So how does Siri work? Speech recognition is at the heart of Siri, but it does not explain the detailed process that Siri understands what users are saying. This week, http://www.aliyun.com/zixun/aggregation/32834.html "> Insiders wrote to explain Siri's work."

When the user speaks to the iphone, the voice will be encoded immediately into a compressed digital file containing all the useful information. This information will be sent to the cloud server via the Internet Service Provider (ISP) network, and the modules in the server will identify what the user is saying.

At the same time, the user's voice will be recognized on the phone side. The speech recognizer installed on the phone will communicate with the cloud server to see if the instructions are suitable for local processing. The user's instructions may require a phone to play a song, while others may require a mobile phone to connect to the network for further assistance. If the speech recognizer believes that the phone's internal module is sufficient to handle the user's instructions, it will tell the cloud server that no longer needs server support.

Depending on the tone and word order of the user, the server will make a static comparison of the speech to see which letters are included in the speech. At the same time, the local speech recognizer will also statically contrast the user's voice. On both the server side and the mobile phone, the most likely identification content will be processed first.

At this point, the identified content already contains a series of vowel and consonant letters. The content is then sent to a language module to evaluate which words are included in the user's voice. Depending on the trustworthiness, the computer creates a list of what the user says.

If the results are credible enough, the computer will be able to understand what the user is saying, such as sending a text message or finding a contact in a contact list. The user will then see what is needed on the phone screen without having to do it manually. In this process, if the user's voice meaning is too vague, the computer will ask the user, such as Erica Orsen (Erica olssen) or Erica Schmidt (Erica Schmidt) that the user wants to find. (Wen/Sina Technology)

(Responsible editor: The good of the Legacy)