TinyML Detects Baby Cry, Triggers Bottle
A developer built a TinyML system using Edge Impulse to detect baby cries and automatically activate a bottle preparation machine, deploying the model to a Portenta H7 via the Arduino library. The workflow uses MFCC preprocessing and a 1D-CNN trained for 100 epochs with a 0.005 learning rate, reporting 98.2% training accuracy and 100% test accuracy. The device enforces a two-hour activation window to avoid repeat activations.
Key Points
- 1Implements tinyML audio classifier using Edge Impulse and Portenta H7 for baby-cry detection
- 2Achieves reported 98.2% training accuracy and 100% test accuracy, indicating strong model performance
- 3Automates bottle preparation via GPIO-controlled machine with two-hour window to prevent repeat activations
Scoring Rationale
Practical, reproducible TinyML tutorial with hardware deployment demonstrates real-world use, but offers limited novelty and single-source validation.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
