Command cards, BACKGROUND card, Audio tally
Open mission materials →
Concept art made with Gemini 3.1 Flash Image← Mission map
Mission 4 of 12Teachable Machine50 to 65 minutes
Wizard School: Voice Spell Rescue
Mochi is trapped in a shifting library. Pixel needs ears that recognize GLOW, HOLD, and ordinary room sound.
Today's AI power
Audio classification
An audio model learns short game commands and the background sounds it should ignore.
Start here
Light three rune paths with separate command and background tests.
Both commands and BACKGROUND become the top label in separate tests.
Frozen Audio Model v1 and Pixel's listening form
See how this game works
Game loop preview
See the power before you build it.
- 1SignalVoice, sound, or room noise
- 2AI moveThe model guesses a sound class
- 3Your ruleThe exact game logic checks what happens next.
- 4Game stateACT
Your mission
One step at a time. One AI power to unlock.
Step 1 of 9Story
Story
Meet the mission problem.
Story signal: Mochi is trapped in a shifting library. Pixel needs ears that recognize GLOW, HOLD, and ordinary room sound.
Do this now
- 1LOOK
Spot the problem in the opening story and tell your crew what needs to change.
Goal
Know what winning looks like.
Your choice: Choose two short commands that sound clearly different.
Watch this change: A clear command activates while ordinary room sound stays BACKGROUND.
Do this now
- 2CHOOSE
Win by proving this: Both commands and BACKGROUND become the top label in separate tests.
Get ready
Gather the props and get your grown-up guide.
Do this now
- 3CHECK
Open the Mission 4 materials on screen or print them, then gather the listed props.
- 4CHECK
Record short game commands and ordinary room sound, then turn the microphone off when testing ends.
Predict
Guess first, then test.
Do this now
- 5CHOOSE
Listen to one second of normal room sound before recording commands.
Build
Make the AI power in the Google tool.
Do this now
- 6MAKE
Follow the Google tool card below. Complete each step in order, then bring the visible result back here.
Google tool step
Open the tool, make the power, then bring one result back.
OPEN TOGETHER
Teachable Machine
Adult and child operate together
Build thisAudio project
Save asAudio Model v1
Do these steps
- Open an Audio Project.
- Create both command classes and BACKGROUND.
- Record balanced one-second examples.
- Train the audio model.
- Test each class.
- Export the model locally and stop microphone access.
You are done whenThe live audio bars separately recognize both commands and ordinary room sound.
Need setup help?
- Bring
- Two command cards
- BACKGROUND card
- Balanced Audio Tally
- Next move
- Record the local version identifier in the passport and freeze it for Missions 9 and 12.
- If stuck
- Shorten similar commands, add normal BACKGROUND examples, and retry without raising your voice.
Bring it back
Return with the result you can see.
Do this now
- 7CHECK
Record its folder name or hash and promise not to retrain it before Missions 9 and 12. Save the exact mission artifact as Audio Model v1, then return to this card.
Glitch test
Make a funny mistake on purpose.
Funny glitch: A harmless book stack performs the spell and bows dramatically.
Do this now
- 8TEST
Test a command while a harmless fan or quiet conversation plays. Notice whether BACKGROUND examples help.
Boss round
Change one thing and retry.
Game move: Voice-powered rescue maze with bouncing-book reactions
Harder remix: Compare near and far tests and explain how a one-second audio window changes the input.
Do this now
- 9REMIX
Choose one evidence-based change, run the same test again, and compare what changed.
- 10TEST
Run the boss round: Audio Model v1 is exported and its version identifier is saved.
Explain
Say what the AI did and prove it.
Power reward: Background Listener · Frozen Audio Model v1 and Pixel's listening form
An audio model learns short game commands and the background sounds it should ignore.
Voice systems must handle silence and room noise, not only perfect commands spoken close to a microphone.
Do this now
- 11SAY
Why does an audio model need examples of ordinary room noise?
- 12SAVE
Complete the mission proof and save only the named artifact: Audio Model v1.
What counts as finished?
Why does an audio model need examples of ordinary room noise?
- Both commands and BACKGROUND become the top label in separate tests.
- The three class counts are recorded.
- Audio Model v1 is exported and its version identifier is saved.
Need help or want a harder challenge?
Make this step easier
Use picture command cards and collect examples in three short rounds with an adult counting.
Say it this way: The AI noticed ___. The exact game rule ___. I know because ___.
Try the deeper remix
Compare near and far tests and explain how a one-second audio window changes the input.
- Record the inputs and outputs.
- Compare two versions or conditions.
- Name the constraint or exact rule.
- Design one fair remix test.
Everyone proves the same AI idea: Both commands and BACKGROUND become the top label in separate tests. · The three class counts are recorded. · Audio Model v1 is exported and its version identifier is saved.
Quick power check