def get_audio_stats(self, data, sample_size):
wf = wave.open(self.WAVE_OUTPUT_FILENAME, 'wb')
wf.setnchannels(self.CHANNELS)
wf.setsampwidth(sample_size)
wf.setframerate(self.RATE)
wf.writeframes(data)
wf.close()
recording_data_dict = sox.file_info.stat(self.WAVE_OUTPUT_FILENAME)
recording_data = list(recording_data_dict.values())
recording_date = datetime.now()
doa = DOA(rate=self.RATE)
doa.put(data)
return {
"MinimumAmplitude": str(recording_data[0]),
"RoughFrequency": str(recording_data[1]),
"ScaledBy": str(recording_data[2]),
"MeanNorm": str(recording_data[3]),
"MaximumDelta": str(recording_data[4]),
"RmsAmplitude": str(recording_data[5]),
"Length": str(recording_data[6]),
"SamplesRead": str(recording_data[7]),
"VolumeAdjustment": str(recording_data[8]),
"MinimumDelta": str(recording_data[9]),
"MidlineAmplitude": str(recording_data[10]),
"RmsDelta": str(recording_data[11]),
"MeanAmplitude": str(recording_data[12]),
"MeanDelta": str(recording_data[13]),
"MaximumAmplitude": str(recording_data[14]),
"DeviceId": self.config.device,
"Direction": doa.get_direction(),
"RecordingDate": recording_date.strftime("%Y%m%d%H%M%S")
}
def main():
src = Source(rate=16000, channels=4, frames_size=320)
ch0 = ChannelPicker(channels=4, pick=0)
kws = KWS(model='snowboy', sensitivity=0.6, verbose=True)
doa = DOA(rate=16000)
src.link(ch0)
ch0.link(kws)
src.link(doa)
def on_detected(keyword):
position = doa.get_direction()
pixels.wakeup(position)
print('detected {} at direction {}'.format(keyword, position))
kws.set_callback(on_detected)
src.recursive_start()
while True:
try:
time.sleep(1)
except KeyboardInterrupt:
break
src.recursive_stop()
def main():
src = Source(rate=16000, channels=4)
ch0 = ChannelPicker(channels=src.channels, pick=0)
kws = KWS(model='snowboy', sensitivity=0.6, verbose=True)
doa = DOA(rate=16000)
src.link(ch0)
ch0.link(kws)
src.link(doa)
def on_detected(keyword):
print('detected {} at direction {}'.format(keyword,
doa.get_direction()))
kws.set_callback(on_detected)
src.recursive_start()
while True:
try:
time.sleep(1)
except KeyboardInterrupt:
break
src.recursive_stop()
# wait a second to allow other threads to exit
time.sleep(1)
def main():
src = Source(rate=16000, frames_size=320, channels=4)
ds = DelaySum(channels=4,
frames_size=src.frames_size,
max_offset=max_offset)
kws = KWS()
doa = DOA(rate=16000, chunks=20)
src.link(ds)
ds.link(kws)
src.link(doa)
def on_detected(keyword):
direction = doa.get_direction()
pixel_ring.wakeup(direction)
print('detected {} at direction {}'.format(keyword, direction))
kws.on_detected = on_detected
src.recursive_start()
while True:
try:
time.sleep(1)
except KeyboardInterrupt:
print('quit')
break
src.recursive_stop()
def main():
src = Source(rate=16000, channels=4)
ch1 = ChannelPicker(channels=4, pick=1)
kws = KWS()
doa = DOA(rate=16000)
src.link(ch1)
ch1.link(kws)
src.link(doa)
outf = "./log/test.wav"
fsall = FileSink(outf, rate=16000, channels=4)
src.link(fsall)
#ch1.link(fs)
for i in range(4):
ch = ChannelPicker(channels=4, pick=i)
src.link(ch)
outf = "./log/test"+str(i)+".wav"
fs = FileSink(outf, rate=16000, channels=1)
ch.link(fs)
def on_detected(keyword):
print('detected {} at direction {}'.format(keyword, doa.get_direction()))
kws.set_callback(on_detected)
src.recursive_start()
sec = 0
while sec < 5:
try:
time.sleep(1)
sec = sec + 1
except KeyboardInterrupt:
break
src.recursive_stop()
def main():
doa = None
# DoA
# src = Source(rate=16000, frames_size=320, channels=4, device_index=0)
src = Source(rate=16000, frames_size=320, channels=4)
doa = DOA(rate=16000, chunks=20)
src.link(doa)
src.recursive_start()
# GPIO
GPIO.output(led_gnd_pin, False)
GPIO.output(led1_pin, True)
GPIO.output(led2_pin, True)
# Google Assistant
parser = argparse.ArgumentParser(
formatter_class=argparse.RawTextHelpFormatter)
parser.add_argument('--credentials',
type=existing_file,
metavar='OAUTH2_CREDENTIALS_FILE',
default=os.path.join(
os.path.expanduser('/home/pi/.config'),
'google-oauthlib-tool', 'credentials.json'),
help='Path to store and read OAuth2 credentials')
args = parser.parse_args()
with open(args.credentials, 'r') as f:
credentials = google.oauth2.credentials.Credentials(token=None,
**json.load(f))
with Assistant(credentials, "RPi3_Assistant") as assistant:
for event in assistant.start():
process_event(event, doa)
def main():
src = Source(rate=16000, channels=4, frames_size=320)
ch1 = ChannelPicker(channels=4, pick=1)
kws = KWS()
doa = DOA(rate=16000)
src.link(ch1)
ch1.link(kws)
src.link(doa)
def on_detected(keyword):
position = doa.get_direction()
pixels.wakeup(position)
print('detected {} at direction {}'.format(keyword, position))
kws.set_callback(on_detected)
src.recursive_start()
while True:
try:
time.sleep(1)
except KeyboardInterrupt:
print('STOP')
break
src.recursive_stop()
def main():
## RESPEAKER_RATE = 16000
## RESPEAKER_CHANNELS = 1
## RESPEAKER_WIDTH = 2
## CHUNK = 1024
## RECORD_SECONDS = 2
src = Source(rate=16000, channels=4, frames_size=320)
ch1 = ChannelPicker(channels=4, pick=1)
kws = KWS()
doa = DOA(rate=16000)
src.link(ch1)
ch1.link(kws)
src.link(doa)
def on_detected(keyword):
position = doa.get_direction()
pixels.wakeup(position)
print('detected {} at direction {}'.format(keyword, position))
src.stream.start()
print("* recording")
frames = []
for i in range(0, int(RESPEAKER_RATE / CHUNK * RECORD_SECONDS)):
data = stream.read(CHUNK)
frames.append(data)
print("* done recording")
src.stream.stop()
print("start to send to baidu")
# audio_data should be raw_data
text = baidu.server_api(generator_list(frames))
if text:
try:
text = json.loads(text)
for t in text['result']:
print(t)
except KeyError:
print("get nothing")
else:
print("get nothing")
kws.set_callback(on_detected)
src.recursive_start()
while True:
try:
time.sleep(1)
except KeyboardInterrupt:
break
src.recursive_stop()
def run(self):
global moving, last_movement_timestamp, doa_valid
src = Source(rate=16000, channels=4, frames_size=320)
#ch1 = ChannelPicker(channels=4, pick=1)
doa = DOA(rate=16000)
#src.link(ch1)
src.link(doa)
src.recursive_start()
self.running = True
while self.running:
try:
time.sleep(1)
current_timestamp = datetime.datetime.now()
if doa_valid == True and (
(current_timestamp - last_movement_timestamp).seconds > 2):
position, amplitute = doa.get_direction()
if amplitute > 2000:
pixels.wakeup(position)
print amplitute, position
if position > 0 and position < 180:
pivot_right()
time.sleep(position / 200)
stop()
elif position >= 180 and position < 360:
pivot_left()
position = 360 - position
time.sleep(position / 200)
stop()
time.sleep(2)
else:
pixels.speak()
else:
pixels.think()
except:
print sys.exc_info()
src.recursive_stop()
def t3_bot():
src = Source(rate=16000, channels=4, frames_size=320)
ch1 = ChannelPicker(channels=4, pick=1)
kws = KWS()
doa = DOA(rate=16000)
src.link(ch1)
ch1.link(kws)
src.link(doa)
pixels.listen()
pwm.setPWM(0, 0, 370)
pwm.setPWM(1, 0, 640)
counter = 0
def on_detected(keyword):
position = doa.get_direction()
pixels.wakeup(position)
print('detected {} at direction {}'.format(keyword, position))
if position >= 30 and position <= 180:
pwm.setPWM(0, 0, 175)
pwm.setPWM(1, 0, 500)
elif position > 180 and position <= 330:
pwm.setPWM(0, 0, 560)
pwm.setPWM(1, 0, 500)
elif position > 330 or position < 30:
pwm.setPWM(0, 0, 370)
pwm.setPWM(1, 0, 6200)
else:
pwm.setPWM(0, 0, 370)
pwm.setPWM(1, 0, 640)
#talkassist.os.system("espeak 'may i help you'")
print("How may I help you?")
print("call google assistant here, delete this line.")
kws.set_callback(on_detected)
src.recursive_start()
while True:
try:
time.sleep(1)
counter += 1
print("counter is at " + str(counter))
except KeyboardInterrupt:
break
src.recursive_stop()
def main():
src = Source(rate=16000, frames_size=160, channels=4)
ch0 = ChannelPicker(channels=src.channels, pick=0)
ns = NS(rate=src.rate, channels=1)
kws = KWS()
doa = DOA(rate=16000, chunks=50)
alexa = Alexa()
alexa.state_listener.on_listening = pixel_ring.listen
alexa.state_listener.on_thinking = pixel_ring.think
alexa.state_listener.on_speaking = pixel_ring.speak
alexa.state_listener.on_finished = pixel_ring.off
# data flow between elements
# ---------------------------
# src -> ns -> kws -> alexa
# \
# doa
src.pipeline(ch0, ns, kws, alexa)
src.link(doa)
def on_detected(keyword):
direction = doa.get_direction()
print('detected {} at direction {}'.format(keyword, direction))
alexa.listen()
pixel_ring.wakeup(direction)
kws.on_detected = on_detected
is_quit = []
def signal_handler(sig, frame):
is_quit.append(True)
print('quit')
signal.signal(signal.SIGINT, signal_handler)
src.pipeline_start()
while not is_quit:
time.sleep(1)
src.pipeline_stop()
pixel_ring.off()
# wait a second to allow other threads to exit
time.sleep(1)
def main():
ser = serial.Serial("/dev/ttyACM0", 9600)
src = Source(rate=16000, channels=4, frames_size=320)
ch1 = ChannelPicker(channels=4, pick=1)
kws = KWS(model='dad.pmdl')
doa = DOA(rate=16000)
src.link(ch1)
ch1.link(kws)
src.link(doa)
def on_detected(keyword):
position = doa.get_direction()
pixels.wakeup(position)
print('detected {} at direction {}'.format(keyword, position))
if position <= 90:
ser.write(b"1000\n")
elif position > 90 and position <= 180:
ser.write(b"0100\n")
elif position > 180 and position <= 270:
ser.write(b"0010\n")
else:
ser.write(b"0001\n")
vibrate_time = 0.5
time.sleep(vibrate_time)
ser.write(b"0000\n")
kws.set_callback(on_detected)
src.recursive_start()
while True:
try:
time.sleep(1)
except KeyboardInterrupt:
break
src.recursive_stop()
ser.close()
def main():
logging.basicConfig(level=logging.DEBUG)
src = Source(rate=16000, channels=4)
ch1 = ChannelPicker(channels=4, pick=1)
ns = NS(rate=16000, channels=1)
kws = KWS(model='snowboy')
doa = DOA(rate=16000)
alexa = Alexa()
alexa.state_listener.on_listening = pixels.listen
alexa.state_listener.on_thinking = pixels.think
alexa.state_listener.on_speaking = pixels.speak
alexa.state_listener.on_finished = pixels.off
def on_detected(keyword):
direction = doa.get_direction()
logging.info('detected {} at direction {}'.format(keyword, direction))
pixels.wakeup(direction)
alexa.listen()
kws.on_detected = on_detected
src.link(ch1)
ch1.link(ns)
ns.link(kws)
kws.link(alexa)
src.link(doa)
src.recursive_start()
while True:
try:
time.sleep(1)
except KeyboardInterrupt:
break
src.recursive_stop()
class XYRobot(object):
jike = config.get('jike', False)
src = Source(rate=16000, channels=4, frames_size=320)
ch1 = ChannelPicker(channels=4, pick=1)
#这里需要填写pmdl的绝对路径
#kws = KWS()
#kws = KWS(os.path.join(constants.DATA_PATH ,config.get('kws_file_name','')))
kws = KWS(
os.path.join(constants.DATA_PATH, config.get('kws_file_name', '')))
doa = DOA(rate=16000)
def __init__(self):
print('极客模式:' + str(self.jike))
self.src.link(self.ch1)
self.ch1.link(self.kws)
self.src.link(self.doa)
def on_detected(keyword):
position = self.doa.get_direction()
pixels.wakeup(position)
voice = os.path.join(
constants.DATA_PATH, 'sysvoices',
'sysvoice' + str(random.randint(1, 8)) + '.mp3')
direction = pixels.positionToDirection()
print('detected {} at direction {} is {}'.format(
keyword, position, direction))
#这里是唤醒后进行打招呼的语音播放
pixels.speak(voice)
print(str(keyword) + voice)
record.speech('你在我' + direction)
if self.jike:
print('进入极客模式')
# 创建线程01，不指定参数
thread_01 = Thread(target=instructions.jikeThread,
args=(
"jikeThread01",
record,
pixels,
))
# 启动线程01
thread_01.start()
else:
#开始录音
outputtext = record.record()
#这里是指令，语音识别后返回文字，将文字传入instructionsFunc，进行处理和播放
instructions.instructionsFunc(outputtext, record, pixels)
self.kws.set_callback(on_detected)
self.src.recursive_start()
config.init()
def sigint_handler(self, signum, frame):
record.stream.stop_stream()
record.stream.close()
record.p.terminate()
self.src.recursive_stop()
print 'catched interrupt signal!'
sys.exit(0)
def train(self, w1, w2, w3, m):
'''
传入三个wav文件，生成snowboy的.pmdl模型
'''
def get_wave(fname):
with open(fname, 'rb') as infile:
return base64.b64encode(infile.read()).decode('utf-8')
url = 'https://snowboy.kitt.ai/api/v1/train/'
data = {
"name":
"wukong-robot",
"language":
"zh",
"token":
config.get('snowboy_token', ''),
"voice_samples": [{
"wave": get_wave(w1)
}, {
"wave": get_wave(w2)
}, {
"wave": get_wave(w3)
}]
}
response = requests.post(url, json=data)
print(response.ok)
if response.ok:
with open(m, "wb") as outfile:
outfile.write(response.content)
return 'Snowboy模型已保存至{}'.format(m)
else:
return "Snowboy模型生成失败，原因:{}".format(response.text)