def capture(self, vad_throwaway_frames=VAD_THROWAWAY_FRAMES):
audio = ""
pcm = alsaaudio.PCM(alsaaudio.PCM_CAPTURE, alsaaudio.PCM_NORMAL,
self.config['device'])
pcm.setchannels(1)
pcm.setrate(VAD_SAMPLERATE)
pcm.setformat(alsaaudio.PCM_FORMAT_S16_LE)
pcm.setperiodsize(VAD_PERIOD)
vad = webrtcvad.Vad(2)
silenceRun = 0
numSilenceRuns = 0
thresholdSilenceMet = False
frames = 0
start = time.time()
events.fire('capture_started')
# do not count first 10 frames when doing VAD
while frames < vad_throwaway_frames:
length, data = pcm.read()
frames = frames + 1
if length:
audio += data
# now do VAD
while ((thresholdSilenceMet is False)
and ((time.time() - start) < self.config['timeout'])):
length, data = pcm.read()
if length:
audio += data
if length == VAD_PERIOD:
isSpeech = vad.is_speech(data, VAD_SAMPLERATE)
if not isSpeech:
silenceRun = silenceRun + 1
# print "0"
else:
silenceRun = 0
numSilenceRuns = numSilenceRuns + 1
# print "1"
# only count silence runs after the first one
# (allow user to speak for total of max recording length if they haven't said anything yet)
if (numSilenceRuns != 0) and (
(silenceRun * VAD_FRAME_MS) > VAD_SILENCE_TIMEOUT):
thresholdSilenceMet = True
path = TMP_PATH + str(uuid.uuid4()) + '.wav'
with open(path, 'w') as rf:
rf.write(audio)
pcm.close()
events.fire('capture_fullfilled', audio=path)
def capture(self, type, *args, **kwargs):
if kwargs['id'] != self.config['trigger']:
return
if not self.config['silent']:
events.fire('speech_requested', audio=RES_PATH + 'alexayes.mp3')
events.fire('capture_requested')
def boot(self, type, *args, **kwargs):
# sys.stdout.write("Trying to reach Amazon...")
# while not check_network():
# sys.stdout.write(".")
# print("")
# print("Connection OK")
if not self._get_token():
sys.exit()
events.fire('greetings_requested')
events.fire('detection_requested')
action="store_true",
default=False,
help="start without saying hello")
parser.add_option('-d', '--debug',
dest="debug",
action="store_true",
default=False,
help="display debug messages")
cmdopts, cmdargs = parser.parse_args()
config['debug'] = True # cmdopts.debug
config['silent'] = cmdopts.silent
# debug logger
def log(type, *args, **kwargs):
print('> {} {}'.format(type, kwargs))
# wiring logger to pubsub
events.register('*', log)
# base components
avs = AVS(config['avs'])
microphone = Microphone(config['microphone'])
synthetizer = Synthetizer(config['synthetizer'])
# third party components
# Ignition
events.fire('boot_requested')
def detect(self):
# create decoders on the fly
if not self.decoders:
self.decoders = []
for id, phrase in self.config['triggers'].iteritems():
config = Decoder.default_config()
# set recognition model to US
config.set_string('-hmm',
os.path.join(get_model_path(), 'en-us'))
config.set_string(
'-dict',
os.path.join(get_model_path(), 'cmudict-en-us.dict'))
# specify recognition key phrase
config.set_string('-keyphrase', phrase)
config.set_float('-kws_threshold', 1e-5)
# hide the VERY verbose logging information
# if not self.config['debug']:
config.set_string('-logfn', '/dev/null')
decoder = Decoder(config)
decoder.id = id
self.decoders.append(decoder)
events.fire('detection_started')
# start decoding
for decoder in self.decoders:
decoder.start_utt()
pcm = alsaaudio.PCM(alsaaudio.PCM_CAPTURE, alsaaudio.PCM_NORMAL,
self.config['device'])
pcm.setchannels(1)
pcm.setrate(16000)
pcm.setformat(alsaaudio.PCM_FORMAT_S16_LE)
pcm.setperiodsize(1024)
phrase = None
triggered = False
while not triggered:
_, buffer = pcm.read()
for decoder in self.decoders:
decoder.process_raw(buffer, False, False)
triggered = decoder.hyp() is not None
if triggered:
phrase = decoder.id
break
pcm.close()
pcm = None
for decoder in self.decoders:
decoder.end_utt()
events.fire('detection_fullfilled', id=phrase)
def process(self, type, *args, **kwargs):
# create request
url = 'https://access-alexa-na.amazon.com/v1/avs/speechrecognizer/recognize'
headers = {'Authorization': 'Bearer %s' % self._get_token()}
payload = {
"messageHeader": {
"deviceContext": [{
"name": "playbackState",
"namespace": "AudioPlayer",
"payload": {
"streamId": "",
"offsetInMilliseconds": "0",
"playerActivity": "IDLE"
}
}]
},
"messageBody": {
"profile": "alexa-close-talk",
"locale": "en-us",
"format": "audio/L16; rate=16000; channels=1"
}
}
with open(kwargs['audio']) as audio:
files = [('file', ('request', json.dumps(payload),
'application/json; charset=UTF-8')),
('file', ('audio', audio,
'audio/L16; rate=16000; channels=1'))]
# send request
response = requests.post(url, headers=headers, files=files)
# cleanup right after use
os.remove(kwargs['audio'])
print(response)
if response.status_code == 204:
events.fire('detection_requested')
return
if response.status_code != 200:
events.fire('error_report_requested', response=response)
return
wrapper = "Content-Type: " + response.headers[
'content-type'] + '\r\n\r\n' + response.content
message = email.message_from_string(wrapper)
for payload in message.get_payload():
content_type = payload.get_content_type()
response_body = payload.get_payload()
if content_type == "audio/mpeg":
filename = TMP_PATH + payload.get('Content-ID').strip(
"<>") + ".mp3"
with open(filename, 'wb') as audio:
audio.write(payload.get_payload())
else:
if content_type == "application/json":
data = json.loads(response_body)
elif self.config['debug']:
print('-- Unknown data returned:')
print(json.dumps(data))
# process audio items first
# if 'audioItem' in data['messageBody']:
# self.player.play_playlist(data['messageBody'])
# pass
# for lisibility
directives = data['messageBody']['directives']
if not directives or len(directives) == 0:
events.fire('detection_requested')
return
wishes = []
for directive in directives:
# speaker control such as volume or mute
if directive['namespace'] == "Speaker":
if directive['name'] == 'SetVolume':
wishes.append({
'type':
'volume',
'value':
int(directive['payload']['volume']),
'relative':
directive['payload']['adjustmentType'] == 'relative'
})
elif directive['name'] == 'SetMute':
pass
# if need of a new capture phase
elif directive['namespace'] == 'SpeechRecognizer' and directive[
'name'] == 'listen':
events.fire('capture_requested',
vad_throwaway_frames=directive['payload']
['timeoutIntervalInMillis'] / 116)
# play speech
elif directive['namespace'] == 'SpeechSynthesizer':
if directive['name'] == 'speak':
wishes.append({
'type':
'speech',
'value':
mrl_fix("file://" + TMP_PATH + directive['payload']
['audioContent'].lstrip("cid:") + ".mp3")
})
# play music
elif directive['namespace'] == 'AudioPlayer':
if directive['name'] == 'play':
pass
for wish in wishes:
if wish['type'] == 'speech':
events.fire('speech_requested', audio=wish['value'])
time.sleep(.1)
events.fire('detection_requested')
def greet(self, type, *args, **kwargs):
events.fire('speech_requested', audio=RES_PATH + 'hello.mp3')
def _callback(self):
events.fire('speech_fullfilled')
