def match(self, data_fft) :
min_distance = da.distance(self.chords[0].chord_fft, data_fft)
best_match = self.chords[0]
for chord in self.chords :
current_distance = da.distance(chord.chord_fft, data_fft)
if current_distance < min_distance :
best_match = chord
min_distance = current_distance
return best_match, min_distance
def learn_from_wave_file(self, file_path) :
print "Learning for "+ self.name
wf = wave.open(file_path, 'rb')
dictionaries = []
p = pyaudio.PyAudio()
stream = p.open(format=p.get_format_from_width(wf.getsampwidth()),
channels=wf.getnchannels(),
rate=wf.getframerate(),
output=True)
data = wf.readframes(var.CHUNK)
while data != '':
dictionaries.append(da.normalize(da.getFft(data, var.RATE)))
data = wf.readframes(var.CHUNK)
stream.stop_stream()
stream.close()
# Average the dictionary
mean_dictionnary = da.average_dictionary(dictionaries)
print "Done Learning"
threshold = 0 ;
for fft_data in dictionaries :
threshold += da.distance(fft_data, mean_dictionnary)
threshold /= len(dictionaries)
print "Done setting the threshold at " + str(threshold)
real_list = []
for fft_data in dictionaries :
if da.distance(fft_data, mean_dictionnary) < threshold :
real_list.append(fft_data)
new_mean = da.average_dictionary(real_list)
print "Done filtering"
print "Distance between the new and old means is " + str(da.distance(mean_dictionnary, new_mean))
threshold = 0 ;
for fft_data in real_list :
threshold += da.distance(fft_data, new_mean)
threshold /= len(real_list)
print "Done resetting the threshold at " + str(threshold)
print len(real_list)
self.chord_fft = new_mean
self.threshold = threshold
