Think Together · Start with a Short Story
The girl says, “Open the story.” The robot first receives a sound signal, not Chinese characters.
Lay Out the Process
- CaptureTurn sound into numbers
- RecognizeEstimate which words were spoken
- Understand the taskWork out the request
- Act or answerGive a result
Speech systems turn sound into data that can be calculated, then use models to recognize or process language.
“Hearing sound,” “recognizing words,” and “understanding what to do” are different steps. Recognition may change in noisy places or with words that sound alike.
Some systems combine these steps, but their output is still not the same as directly reading a person's mind.
Now You Try
If “fifteen” is recognized as “fifty,” what might happen to a later calculation?
Ready? Open this to see the explanation
The input number has changed. Even if all later arithmetic is correct, the system has answered the wrong question. Check the recognized words first.
One Idea to Take Away
Speech systems turn sound into data that can be calculated, then use models to recognize or process language.
Try covering the diagram above and explaining each step to someone in your family. If one step is hard to explain, return to that card and look again.