micro:bit + Teachable Machine

1. Introduce the activity 
   • Discuss
     1. What if your computer could hear the difference between a clap and a whistle?
     2. How would we teach a computer to recognize sounds?
   • Today, you’re going to train your own AI. 
2.  Introduce Machine Learning
   • Machine learning means teaching a computer using examples instead of step-by-step instructions.
   • Explain simply:
     1. You give examples (dataset)
     2. The computer finds patterns in the data
     3. It predicts what new inputs are
   • We’re going to teach it three sounds: snap, clap, and whistle.
3. Train the Model
   • Guide students in Teachable Machine:
     1. Start a Sound Project
     2. Create classes:
        1. Snap
        2. Clap
        3. Whistle
     3. Record multiple samples for each
   • The more examples you give, the smarter your AI becomes. Models learn by using labeled examples of each class
4. Train & Test
   • Now let’s see what your AI learned.
   • Students:
     1. Click Train Model
     2. Test each sound
   • Discuss:
     1. Does it recognize your sounds correctly? 
   • Optional:
     1. If it does not recognize your sounds add more data to each class and retrain.  
5. Connect to micro:bit
   • Now we connect your AI to a real device. We will use the Teachable Micro:bit App, the app sends the detected sound label to your micro:bit.
   • Guide students to connecting teachable machine to the Teachable Micro:bit App:
     1. Export the Model to the Cloud – Within Teachable Machine, export the model to the cloud. You will receive a link that looks something like this: https://teachablemachine.withgoogle.com/models/Jgyr546VH/ 
     2. Open the Teachable Micro:Bit App – On your laptop or mobile device, open the app and click the START image button. A prompt window will appear where you need to paste the link to your Teachable Machine model.
     3. Allow Camera Access – Allow the browser to use your device camera. After loading the model, you should see a camera preview and the names of detected classes displayed below the camera window.
     4. Prepare Your Micro:Bit – Use MakeCode to write a program that allows the Micro:Bit to connect to Bluetooth UART and receive string data. You can find a sample code snippet below.
     5. Connect to the Micro:Bit – Click the robot head icon in the upper-right corner to connect to the Micro:Bit board.
     6. Control Actions – Once the connection is established, the Micro:Bit will receive class names from the app. You can then control the robot’s actions based on these messages using your Micro:Bit code.  
     7. For help see link: https://cardboard.lofirobot.com/teachable-microbit-app-info/
6.  Program micro:bit Responses
   • Use MakeCode to program the micro:bit. The program will activate bluetooth UART on the Micro:Bit board and display received strings as text on the board display.
   • In MakeCode:
   • Use Bluetooth UART (receive string)
   • Add logic:
     1. If message = “clap” → show icon
     2. If message = “snap” → play sound
     3. If message = “whistle” → display animation
   • Walk students through creating the following code or use the provided link to share the completed code with students.
     1. 
7. Test the System
   • Download the code to your micro:bit
   • Students:
     1. Snap
     2. Clap
     3. Whistle
   • Discuss
     1. Does the micro:bit respond correctly?
8. Improve the Model
   • If it’s not working perfectly—that’s normal. We can update the dataset to improve our model. 
   • Guide improvements:
     1. Add more training samples to each dataset
     2. Clean / Remove inconsistent data
     3. Retrain model

• The micro:bit must be worn for this activity
• The micro:bit doesn’t collect data on sound, so connecting it to AI only uses movement / accelerometer data
• Expect:
  • Inconsistent models (this is GOOD for learning)
  • Need for clear, repeated training data
• Emphasize:
  • AI depends on the data we give it
  • AI makes predictions, not perfect decisions
  • Humans are responsible for training and improving AI

• How did your AI learn to recognize sounds?
• What made your model more accurate?
• What mistakes did it make?
• How is this different from normal coding?

Students can:

• Add more sound categories (e.g., tap, voice commands)
• Control external components (LEDs, servos)
• Build a sound-controlled device
• Compare AI vs non-AI sound detection