AI Tools
AI is a tool. You're the designer.
Today and the next session, you'll generate sound effects with AI and train your own AI model on Spanish vocabulary. Then you'll decide how (and whether) to integrate these tools into your Makey Makey Foley board.
Some sounds are nearly impossible to Foley by hand — a deep engine rumble, a distant explosion echoing across space, a dramatic orchestral sting. AI can generate them in seconds from a text prompt.
You're not replacing your Foley work. You're ADDING ambient and music layers that your physical board can't easily produce.
Sign in. Try a few simple prompts to see how the tool responds:
- "deep space engine rumble, low frequency, ambient"
- "distant battle, explosions echoing"
- "dramatic orchestral sting, brass and strings"
Notice: more specific prompts = better results. "Music" is vague. "Slow piano with reverb in a large empty hall" is useful.
What sounds does your group need for the final performance that you CAN'T Foley by hand?
- Ambient ship bridge — quiet hum, distant control beeps?
- Laser missile incoming?
- Engine roar when Tochiro accelerates?
- A dramatic musical sting for the cat proverb?
Generate at least 3 sounds your group wants to use. Download each one to your device.
Generate the same kind of sound with an English prompt AND a Spanish prompt. Compare. Does the AI respond differently? Translation isn't just for humans — it shapes AI output too.
You're going to TRAIN an AI to recognize Spanish vocabulary you say out loud. No coding, no math — just samples and clicks. Then you can use your trained model however you want.
Watch this first (15 minutes) to see what's possible:
As a group, pick 3 to 5 Spanish words from the Captain Harlock clip you want the AI to recognize. Some strong candidates:
- trasero (the false friend — "rear/behind")
- confiar (the other false friend — "trust")
- cañón (cannon)
- enemigo (enemy)
- timón (helm / ship's wheel)
- Click "Get Started" → "Audio Project"
- Record 20 seconds of "Background Noise" (just silence in your space)
- Create one class per Spanish word — label it with the word
- For each class: record at least 30 samples of the word. Take turns. Different voices = stronger model.
- Click "Train Model" — wait a few seconds
- Say each word — does the AI recognize it? Check the confidence bars.
- Try saying a word someone ELSE in your group recorded — does it still work?
- Try saying a word with the WRONG pronunciation — does it get fooled?
- If the model is weak: go back, record more samples, retrain
You now have three things:
- A Makey Makey Foley board with your own recorded sounds
- AI-generated ambient sounds and music (from Music FX)
- A trained AI that recognizes Spanish words (from Teachable Machine)
For the final performance, you can use any combination:
- Easy: Add AI sounds to your Scratch project. Now your board triggers them too.
- Medium: Use AI music as background ambience, play it from one phone, while your board does Foley.
- Hard (optional): Connect Teachable Machine to Scratch so saying a Spanish word triggers an effect. Real-time AI integration. Ask if you want help with this.