AI Soundwaves Exploration | San Diego Unified School District

In this rapid sprint, you'll use Google's Teachable Machine to build audio-specific machine learning models and then test the limits of what your models can and cannot recognize.

Activity Prompt

WHAT WE ARE MAKING

WHY WE ARE MAKING IT

We will be exploring the how to build a machine learning model that can recognize distinct sounds and what can cause the algorithm to stop functioning properly

To learn more about machine learning models, how they function, and their llimitations

Final Project Requirements

Your project will be a unique machine learning model trained to recognize 3 distinct sounds.

+

؜2nd

3rd

4th

Introduction: Learn about machine learning and audio recognition

Explore Teachable Machine

Train your model

؜Test your model

ACTIVITY STEPS

1st

؜AI Soundwaves Activity

1st

Before we get started exploring machine learning audio models, you'll learn a bit more about how machine learning works and consieder why AI sound recognition may not always be reliable.

First, watch this comedic video about an inherent limitation of voice recognition.

Introduction

Alexa after man goes to dentist video

How do you think AI can recognize and process sounds?

Perhaps the next video can help you understand more about the processes involved...

Discussion

What is Machine Learning?

Machine learning (ML) is a subset of AI that enables computers to learn from data and make decisions or predictions without being explicitly programmed for every rule. By identifying patterns within large datasets, these systems improve their accuracy over time to perform tasks like image recognition, natural language processing, and predictive analytics

؜Watch this video from IBM to learn more!

2nd

؜Go to Google's Teachable Machine: https://teachablemachine.withgoogle.com/

Click Get Started. Then choose Audio Project.

Visit Teachable Machine

3rd

Train your Audio model. Follow the steps on the page. You'll need to record background audio sound so the model can better understand your sonic surroundings. Then choose two other sounds to record. Try to be creative! You can sing, whistle, say a word/phrase.

Train Your Audio Machine Learning Model

4th

؜Now that your model has been trained and works reliably, get with another group and have them test our your audio inputs. How well does it work when another team tries to make your sounds? Discuss!

Test Your Model's Limitation

AI Soundwave Exploration

Or