Siri principle of operation: local voice recognition analysis of cloud services, content

According to ZDNet reported that the "voice recognition" is the iPhone 4S in the heart of Voice Assistant Service Siri, Siri does not explain the contents of said user to understand the detailed process.
Siri is what works?
Insiders this week the author, explained Siri's working procedures.
When the user of the iPhone 4S words, the voice was immediately converted into encoded into a compressed digital file, the file contains all the useful information.
This information will be through the Internet service provider (ISP) network transmission to the cloud computing servers, and server module will identify the user saying.
At the same time, the user's voice will be recognized in the mobile terminal.
Installed in the mobile phone voice recognition device will communicate with the cloud computing server, whether it is suitable for instruction to be processed locally.
The user's command may require mobile phones to play a song, while others are likely to need phone instructions to connect to the network, for further assistance.
If the speech recognizer that mobile phone users within the module is sufficient to handle the instruction, then will inform the cloud computing server, the server no longer needs the support.
According to the user's tone and word order, the server will compare the static voice, to understand what the letter contained in the voice.
At the same time, the speech recognizer local user voice will also be static contrast.
On the server side and mobile side, the possibility of identifying the highest priority content will be dealt with.
At this point, the contents have been identified include a series of vowels and consonants.
Then these will be sent to a language module, to assess what is included in the user's voice in words.
Depending on the credibility, the computer will create a list of the contents of said user.
If you have enough credibility to judge the results, then the computer will be able to understand the contents of said user, for example, send text messages or contact list to find contacts.
Then the user will see the phone screen displays the content, without manual operation.
In this process, if the user's voice too vague, then the computer will ask the user, such as the user wants to find the contact is Eric Olson (Erica Olssen) or Eric Schmidt (Erica
Schmidt).




0
Look at the top