def generateAudio(self, duration, track): if duration == 0: return # generate the base empty sample seconds = (60.0 / (self.bpm * self.ppq)) * duration length = int(seconds * self.sampleRate) sample = numpy.zeros(length) melodysample = numpy.zeros(length) # add in any notes that are active #print "adding some samples in" for i in range(0, 128): if (self.noteActive[i]): #print "note: " + str(i) frequency = midi_util.midiToFrequency(i) #print "adding frequency " + str(frequency) sample += wave_gen.saw(frequency, seconds, self.sampleRate) if track == 0: #melody track for chorales melodysample += wave_gen.sine(frequency, seconds, self.sampleRate) # concatenate this to the output self.tracks[track] = numpy.concatenate([self.tracks[track], sample]) if track == 0: #melody track for chorales self.output[1] = numpy.concatenate([self.output[1], melodysample])
print "error: " + str(new_error)
if abs(error - new_error) < 0.1: break
error = new_error
# save the network
print "Saving neural network..."
NetworkWriter.writeToFile(net, os.path.basename(dirname) + 'net')
if __name__ == '__main__':
dirname = os.path.normpath(sys.argv[1])
# wave_reader.extractFeatures(track)
trainNetwork(dirname)
net = NetworkReader.readFrom(os.path.basename(dirname) + 'net')
# predict on some of the training examples
print "Predicting on training set"
data = numpy.genfromtxt(os.path.join(dirname, 'train09_seg.csv'), delimiter=",")
labels = numpy.genfromtxt(os.path.join(dirname, 'train09REF.txt'), delimiter='\t')[0::10,1]
## for i in range(200):
## print net.activate(data[i]), labels[i]
cdata = numpy.array([])
for feature in data:
freq = max(0, net.activate(feature))
sample = wave_gen.saw(freq, 0.1, 44100)
cdata = numpy.concatenate([cdata, sample])
wave_gen.saveAudioBuffer('test.wav', cdata)
## for freq in labels:
## sample = wave_gen.saw(freq, 0.1, 44100)
## cdata = numpy.concatenate([cdata, sample])
## wave_gen.saveAudioBuffer('test_ref.wav', cdata)
def plotFourierTransform(sample): fourierTrans = numpy.abs(numpy.fft.fft(sample)) fourierTrans = fourierTrans[0:(fourierTrans.size / 2)] fig, ax = plt.subplots() ax.plot(fourierTrans) plt.show() def plotSample(sample): fig, ax = plt.subplots() ax.plot(sample) plt.show() if __name__ == '__main__': # create a sine wave at the desired frequency sineWave = wave_gen.saw(440, 1, 44100) #+ wave_gen.sine(1000, 1, 44100) #sineWave = sineWave + (wave_gen.noise(1, 44100) * 10) # create the plot fig, ax = plt.subplots() # get the fourier transform sineFreq = numpy.abs(numpy.fft.fft(sineWave)) # we only care about the positive frequencies sineFreq = sineFreq[0:(sineFreq.size / 2)] print numpy.max(sineFreq) # plot our fourier transform #saw = wave_gen.saw(440, 1, 44100) print sineWave[0] ax.plot(sineFreq)
#produces a result wav so you can listen to the prediction
if __name__ == '__main__':
dirname = os.path.normpath(sys.argv[1])
#track = os.path.join(dirname, '000106b_.wav')
print "Reading wav into input data..."
data = wave_reader.wavToFeatures('000106b_.wav')
numData = data.shape[0]
labels = numpy.zeros(numData)
print "Reloading neural network..."
net = NetworkReader.readFrom(os.path.basename(dirname) + 'designnet')
print "Activating neural network..."
for i in range(numData):
labels[i] = net.activate(data[i])
print "Generating result wav..."
cdata = numpy.array([])
for label in labels:
#if(freq > 0):
# freq = midi_util.frequencyToNoteFrequency(label)
label = round(label)
freq = midi_util.midiToFrequency(label)
sample = wave_gen.saw(freq, 0.25, 44100)
sample = wave_gen.saw(freq, 0.25, 44100)
cdata = numpy.concatenate([cdata, sample])
print "Saving result wav..."
wave_gen.saveAudioBuffer('000106b_predict.wav', cdata)
def saveFile(filename, data):
numpy.savetxt(filename, data, delimiter=',', fmt='%.2f', newline = '\n')
def addToFeatureMatrix(matrix, feature):
feature = feature.reshape(1, feature.shape[0])
if matrix.size is 0:
matrix = feature;
else:
matrix = numpy.vstack((matrix, feature))
return matrix
if __name__ == '__main__':
matrix = numpy.array([])
seconds = 1
for note in range(0, 12):
print "generating note " + str(note)
for x in range(0, 100):
notesToPlay = chord_gen.createMajorChords(note)
length = int(seconds * 44100)
sample = numpy.zeros(length)
for i in range(0, len(notesToPlay)):
freq = midi_util.midiToFrequency(notesToPlay[i])
sample += wave_gen.saw(freq, seconds, 44100)
features = analyzer.getFrequencies(sample)
features = numpy.append(features[0:5000], note)
#features = features.astype(numpy.int16)
matrix = addToFeatureMatrix(matrix, features)
saveFile('largedata.csv', matrix)
