<aside> 💡 Craft your own interactive Microbit project that responds to your voice!

</aside>

<aside> 👉 Try it out here: https://ttseng.github.io/micropal/

</aside>

<aside> 💻 System Requirements

• Microbit • 1-2 servo motors • Microbit Servo Board (if using more than 1 servo motor) • USB programming cable / batteries

• Laptop or desktop computer running Chrome

Note: App is not mobile or tablet compatible – sorry!

</aside>

<aside> 🌐 Tools Used

• Teachable Machines (for creating audio machine learning model) • MicroPal App (for programming)

</aside>

<aside> ⚠️ Note that this project is actively under development! If you have feedback or suggestions, please share here.

</aside>

An ~5 min walkthrough presented at Microbit Live

Pre-Setup: Test Your Microbit Bluetooth Connection

MicroPal uses Web Bluetooth to connect to your Microbit. To use Web Bluetooth, you must open the website in Chrome on Desktop (mobile and tablet is not supported - sorry!)

You'll first upload a custom Hex file to your Microbit and then pair using the MicroPal website.

Load this Hex file onto your Microbit (download the file and click and drag onto your Microbit)
Open the MicroPal website and click the Pair Microbit button
Once you've paired, test out that you're able to program the LED display – click test after adding a design:

https://s3-us-west-2.amazonaws.com/secure.notion-static.com/3ee7f616-3c10-44c4-8881-5e48de41bacf/DEMO.mp4

Step 1: Creating a Teachable Machine Model

Open Teachable Machines, where you'll build your own audio machine learning model. With machine learning, you teach a computer to recognize different events by providing examples, or data samples.

When designing your own Micropal, think about different sounds or commands you'll want your pet to respond to, such as a greeting (hi, goodbye) or a sound effect (clap, humming a tune).

First, you'll need to record about ~20 seconds of background noise to calibrate the model. Click the Mic button under Background Noise to record and extract a 20s sample.
Now click the Add a Class button to define a new sound you want your MicroPal to recognize. Give it a name, and click Mic to record samples. Each new sound, or class, requires at least 8 samples. After you Record 2 seconds, click Extract Sample to add to your total number of audio samples.
Once you have enough samples for each class, click train model and test out how well it works.

When you repeat any of the sounds you trained, you should see the percentage spike.

Ideas for improving your model
When you're happy with your model, click the Export Model button.
Now, click Upload my Model , which will give you a sharable link you can add to the MicroPal website. Click Copy to copy the URL.

Step 2: Add Model to MicroPal Website

Now you'll be able to program your Microbit using the MicroPal website.

Paste the URL from Teachable Machines into the form at the top of the page and click Load Model
Once your model is loaded, you'll see a form appear for each sound in your model. You'll be able to program how your Microbit's display and servo motors should respond when each sound is recognized.