Skip to content

Model training and evaluation pipeline

Jump to bottom
David J. Klein edited this page Jun 8, 2016 · 8 revisions 
>> annotateWav(wavFile, csvFile); % Did this separately for each WAV file, each appends to .csv
>> annotateKw(dirName, csvFile); 
>> getAnnotatedData('~/Dropbox/Data/keyword','~/Dropbox/Data/keyword/annotations.csv','~/Dropbox/Data/keyword/clipData3.mat');
>> createTrainingWavs('~/Dropbox/Data/keyword/clipData3.mat','~/Dropbox/Data/keyword/trainingWavs')

$ python createTrainingFeatures.py ~/Dropbox/Data/keyword/trainingWavs    
$ python train_spec.py model_may31_small_sigm cnn
$ python predict_spec.py test.wav model_may31_small_sigm
$ python online_detector.py model_may31_small_sigm
Clone this wiki locally