Recognising pictures and speech
AI can recognise what is in a picture and turn spoken words into text. It does this by turning pictures and sounds into numbers and finding patterns it learned from many examples.
📌 Key points
- A digital picture is made of tiny squares called pixels. Each pixel is stored as numbers for its colour.
- An AI that recognises pictures learns from many labelled pictures. It often learns simple patterns first (edges, colours, shapes) and builds up to bigger ones (eyes, wheels, faces).
- This is called image recognition. It is used in photo apps, medical scans and cars that can spot people crossing the road.
- Sound is turned into numbers too. Speech recognition turns spoken words into text.
- Speech recognition learns from many recordings of people talking, with the correct words written down.
- AI can make mistakes with blurry pictures, unusual angles, background noise or accents it did not hear much in training.
- Face recognition uses pictures of people's faces, which is personal data. It must be used carefully and fairly.
📖 Key words
- Pixel
- One tiny square of colour in a digital picture.
- Image recognition
- When AI identifies what is in a picture.
- Speech recognition
- When AI turns spoken words into text.
- Face recognition
- When AI identifies a person from their face.
- Accent
- The way people from a particular place say words.
✏️ Worked example
How does a photo app learn to find pictures of cats?
- It is trained on many thousands of photos labelled "cat" or "not cat".
- To the computer, each photo is just numbers for the colour of each pixel.
- The AI learns patterns in those numbers: edges and shapes first, then things like pointy ears, whiskers and fur.
- Given a new photo, it looks for those patterns and predicts "cat" or "not cat".
- Answer: by learning patterns from many labelled examples. It can still be wrong, for example with a toy cat or a blurry photo.
Try this with a grown-up: use a voice-typing tool and speak clearly, then very fast, then with noise in the room. Notice when it makes mistakes and think about why.
Test yourself8 questions from a bank of 20. A new mix every time.
Start the quiz →