===============================================================================
== Approach #1
===============================================================================

NOTES: 
Approach #1 was the one from the Textbook, chapters 7,8.
Done during last 2 weeks


===============================================================================
== Approach #2: KeyWord Spotting Application: the EdgeImpulse Approach
===============================================================================

Develop a tinyML model to spot keywords “marquette” and “engineering”.
CNN classifier is developed and trained in Edge Impulse (EI).

---Step 1:
Record audio files 1 sec long.
Record 50 "marquette" and 50 "engineering".
Use a third class "silence". Copy 50 files from faucet_dataset/faucet 
(as described in W12_2_Tutorial_KeyWord_Spotting.docx)
The dataset I have created like so is included here as dataset2.zip

---Step 2:
Create EdgeImpulse project and use the above dataset to develop the model;
as described in classm on slides, and in the tutorial.

At the end of the EI project development, Arduino library is downloaded from EI 
and used with Arduino Nano 33 BLE Sense board. 
The download from EI is (2026 version):
  ei-keyword_spotting_project_2026-arduino-1.0.1-impulse-#1.zip

And the unzipped folder is (to look at the code and change):
  keyword_spotting_project_2026_inferencing/

---Step 3:
The application can be run from within Arduino IDE:
--Install lib from .zip. Go to Sketch->Include Library->Add .ZIP Lib
--Then go to Files->Examples... (see PPT slides)
