Experience AI learning by Voice Recognition Kit and neural network

Objective:
To help students experience how neural network can be applied to voice recognition by adopting an open source deep learning framework for on-device machine learning inference, and make use of pre-trained models and ready-made datasets.

Description:
The students can make use of previously configured voice recognition kit and download an open source deep learning framework from the Internet and install to the micro-controller of the kit. Meanwhile, free pre-trained models and sample raw datasets can be downloaded from the framework provider portals or Git repositories.

Hardware
  • Microcontroller boards, intelligent voice board and other accessory kits such as LED bezel, flex cable, Micro SD card, push button, Micro SD card, cardboard
  • Smartphone, Windows / Mac / Linux computer, Wi-Fi connection OR
  • Monitor, Windows / Mac / Linux computer, Mouse, Keyboard, Wi-Fi connection
Software
  • Common programming languages like Python. Linux and neural network framework like Tensorflow Lite.
Cloud Services
  • Nil
Major Activities A more advanced workshop on DIY voice recognition kit. The school will provide the necessary hardware, software and required tools.

  • Enable the network connection with WiFi and obtain IP address.
  • Download and install the updated Tensorflow Lite framework, sample datasets and pre-trained models to the previously configured DIY voice kit.
  • Students can run these models and try out those APIs to see how deep learning work for more complex dialogue.
Learning Objective(s)
  • Understand how neural network runs on-device.
  • Customise pre-trained models with sample dataset.
  • Experience more complex voice recognition.
  • Learn and study the AI sample source code provided
Duration
  • Theoretical session : 1 hour
  • Practical sessions : 30 hours
Difficulty
  • Medium - Hard
Target Level S4 - S6
Target No. of Students Around 3 - 4 students per one voice kit