Malay Command Recognition using ANN

• This project is to recognize the malay command given for electrical appliances

• The electrical appliance choose for this work are fan and light

• The command used are BUKA KIPAS, TUTUP KIPAS, BUKA LAMPU and TUTUP LAMPU

• The detail of this project are explain in the REPORT

• How to used this code:

  • Step 1 : load the command_feature.mat (double click at the mat file)
  • Step 2 : run the ann_classification.m to train and validate the malay command.
    • The ANN code is generate using neural network toolbox.
    • To start the toolbox type nnstart in the command windows and choose Pattern Recognition Apps
    • This is the step how to used nnstart(nprtool)
  • Step 3 : Predict the command

• There are several steps need to follow to recognize the malay command given.

  • First process is pre-processing process. The step for pre-processing are:
    • Choose Signal >> Pre-Emphasis >> Low Pass Filtering

• Second process is feature extraction. The feature extraction used is MFCC. The step are:
  • Frame blocking + Windowing(Hamming Windowing) >> Fourier Transform >> Mel Frequency Wrapping

• Lastly, Classification method using ANN
  • The threshold set for command prection is 0.5 (this threshold can change based on your dataset)
  • This is the prediction example for this signal

• In this project there are some weaknesses need to be modified

  • The feature extraction is done manually and save in mat file to be used in ANN because the melceps give different lenght of output.
  • This is the mat file of MFCC feature extraction
  • Neural network need same data size to be classified
  • Please tag me if the problem can be solve.🥰

• Specification used for this work:

  • Matlab 2014a

⛔ The dataset is not provided to protect personal right

😃 The command example is provided