def LoadAudioData():
ass = iv.ShortVector(1)
n, sr, cc = studio.LoadAudio(model_path + "\\yes.wav", ass)
spectrum = iv.ComplexVector(10)
r = tranform.STFFT(ass, 128, 161, True, spectrum)
aspectrum = studio.ToNumpy(spectrum)
spect = np.absolute(aspectrum)
spect = np.log1p(spect)
spect = spect.reshape(161, -1)
spect = torch.FloatTensor(spect)
mean = spect.mean()
std = spect.std()
spect.add_(-mean)
spect.div_(std)
return spect
def GetAudioData():
ass = iv.ShortVector(1)
studio.GetAudioStream("Record", ass, 5.0, 16000.0, 1)
#studio.WriteAudio("c:\\data\\test.wav", ass, 16000.0, 1)
spectrum = iv.ComplexVector(10)
r = tranform.STFFT(ass, 128, 161, True, spectrum)
aspectrum = studio.ToNumpy(spectrum)
spect = np.absolute(aspectrum)
spect = np.log1p(spect)
spect = spect.reshape(161, -1)
spect = torch.FloatTensor(spect)
mean = spect.mean()
std = spect.std()
spect.add_(-mean)
spect.div_(std)
return spect