Skip to content

Latest commit

 

History

History
43 lines (33 loc) · 1021 Bytes

README.md

File metadata and controls

43 lines (33 loc) · 1021 Bytes
Raw

1. Specific words Recognition, using MATLAB

A. Pre-emphasis
B. Hamming Window
C. Feature Extraction
D. Training
E. Test

2. Open/Close door-lock with Arduino

1. Specific words Recognition, using MATLAB

A. Pre-emphasis:

	1. smoothing filter

	2. remove silence

B. Hamming Window

C. Feature Extraction

	1. pitch
		split Voice and Unvoice
		(we use Voice Only)
	2. MFCC
		13 first MFCC coefficient

D. Training

	1. standardize
	2. KNN algorithm
	
	use makeNewClassifier.m

trainedClassifier is in "func" folder already; Use 'makeNewClassifier.m' if you need train new classifier.

notice: extract trainFiles.zip in order to use for training

E. Test

	use CheckAudio.m

The steps above are roughly taken from this MATLAB page.

2. Open/Close door-lock with Arduino

	upload "/arduino/door/door.ino" to the Arduino board
	use CheckAudioA.m