class VoiceCmd:
def __init__(self, config):
self.config = config
self.commands = {}
self.command_names = {}
commands_folder = self.config["commands_folder"]
for index, command_name in enumerate(os.listdir(commands_folder)):
cmd = Command(command_name, index, os.path.join(commands_folder, command_name))
self.commands[command_name] = cmd
self.command_names[index] = command_name
self.svm = LibSvm(svm_type="c_svc", kernel_type="linear")
x, y = [], []
for command in self.commands.itervalues():
for feature in command.objects:
x.append(feature)
y.append(command.index)
self.svm.learn(x, y)
def run(self):
print "Speak now."
signal = self.read_voice()
# signal = load_wav('commands/calculator/command1.wav')
print "Recording stopped"
decision = self.predict(signal)
print decision
self.commands[decision].execute()
def predict(self, signal):
return self.command_names[int(self.svm.pred(signal))]
def distance(self, template, query):
return dtw(template, query)[0]
def read_voice(self):
rate, fpb, seconds, channels = 44100, 1024, 3, 1
read = []
p = pyaudio.PyAudio()
stream = p.open(format=pyaudio.paInt16, channels=channels, rate=rate, input=True, frames_per_buffer=fpb)
chunks_needed = rate / fpb * seconds
for i in xrange(chunks_needed):
data = stream.read(fpb)
read.append(data)
stream.close()
p.terminate()
data = "".join(read)
debug = wave.open("debug.wav", "wb")
debug.setnchannels(channels)
debug.setsampwidth(p.get_sample_size(pyaudio.paInt16))
debug.setframerate(rate)
debug.writeframes(data)
debug.close()
out = unpack_wav(data, chunks_needed * fpb, channels)
return out
def bench_mlpy(X, y, T, valid):
#
# .. MLPy ..
#
from mlpy import LibSvm
start = datetime.now()
clf = LibSvm(kernel_type='rbf', C=1., gamma=1./sigma)
clf.learn(X, y.astype(np.float64))
score = np.mean(clf.pred(T) == valid)
return score, datetime.now() - start
def bench_mlpy(X, y, T, valid):
#
# .. MLPy ..
#
from mlpy import LibSvm
start = datetime.now()
clf = LibSvm(kernel_type='rbf', C=1., gamma=1. / sigma)
clf.learn(X, y.astype(np.float64))
score = np.mean(clf.pred(T) == valid)
return score, datetime.now() - start
def metric(self):
totalTimer = Timer()
with totalTimer:
model = LibSvm(**self.build_opts)
model.learn(self.data_split[0], self.data_split[1])
if len(self.data) >= 2:
predictions = model.pred(self.data[1])
metric = {}
metric["runtime"] = totalTimer.ElapsedTime()
if len(self.data) == 3:
confusionMatrix = Metrics.ConfusionMatrix(self.data[2],
predictions)
metric['ACC'] = Metrics.AverageAccuracy(confusionMatrix)
metric['MCC'] = Metrics.MCCMultiClass(confusionMatrix)
metric['Precision'] = Metrics.AvgPrecision(confusionMatrix)
metric['Recall'] = Metrics.AvgRecall(confusionMatrix)
metric['MSE'] = Metrics.SimpleMeanSquaredError(
self.data[2], predictions)
return metric
class VoiceCmd:
def __init__(self, config):
self.config = config
self.commands = {}
self.command_names = {}
commands_folder = self.config['commands_folder']
for index, command_name in enumerate(os.listdir(commands_folder)):
cmd = Command(command_name, index,
os.path.join(commands_folder, command_name))
self.commands[command_name] = cmd
self.command_names[index] = command_name
self.svm = LibSvm(svm_type='c_svc', kernel_type='linear')
x, y = [], []
for command in self.commands.itervalues():
for feature in command.objects:
x.append(feature)
y.append(command.index)
self.svm.learn(x, y)
def run(self):
print "Speak now."
signal = self.read_voice()
#signal = load_wav('commands/calculator/command1.wav')
print "Recording stopped"
decision = self.predict(signal)
print decision
self.commands[decision].execute()
def predict(self, signal):
return self.command_names[int(self.svm.pred(signal))]
def distance(self, template, query):
return dtw(template, query)[0]
def read_voice(self):
rate, fpb, seconds, channels = 44100, 1024, 3, 1
read = []
p = pyaudio.PyAudio()
stream = p.open(format=pyaudio.paInt16,
channels=channels,
rate=rate,
input=True,
frames_per_buffer=fpb)
chunks_needed = rate / fpb * seconds
for i in xrange(chunks_needed):
data = stream.read(fpb)
read.append(data)
stream.close()
p.terminate()
data = ''.join(read)
debug = wave.open('debug.wav', 'wb')
debug.setnchannels(channels)
debug.setsampwidth(p.get_sample_size(pyaudio.paInt16))
debug.setframerate(rate)
debug.writeframes(data)
debug.close()
out = unpack_wav(data, chunks_needed * fpb, channels)
return out
