add_utt("(hi|hello|ok) computer", Intent.HELLO)
add_utt("switch (on|off) the (light|lights)", Intent.LIGHT)
add_utt("switch the (light|lights) (on|off)", Intent.LIGHT)
add_utt("switch (on|off) the (music|radio)", Intent.RADIO)
add_utt("switch the (music|radio) (on|off)", Intent.RADIO)
rec.start_recording()
print ("Please speak. (CTRL-C to exit)")
while True:
samples = rec.get_samples()
audio, finalize = vad.process_audio(samples)
if not audio:
continue
user_utt, c = asr.decode(audio, finalize)
print ("\r%s " % user_utt, end='', flush=True)
if finalize:
print ()
best_dist = ED_THRESHOLD
intent = None
for utt in utt_map:
dist = edit_distance (tokenize (utt, lang='en'),
tokenize (user_utt, lang='en'))
if (dist<ED_THRESHOLD) and (dist<best_dist):
best_dist = dist
intent = utt_map[utt]
if intent == Intent.HELLO:
if msg_queue:
data = msg_queue.pop(0)
finally:
msg_cond.release()
if data:
if data['topic'] == TOPIC_INPUT_AUDIO:
audio = data['pcm']
do_finalize = data['final']
loc = data['loc']
logging.debug('asr.decode...')
hstr2, confidence = asr.decode(SAMPLE_RATE,
audio,
do_finalize,
stream_id=loc)
logging.debug('asr.decode...done')
if do_finalize:
logging.info("asr: %9.5f %s" % (confidence, hstr))
if hstr2:
hstr = hstr2
astr = '...'
data = {}
data['lang'] = lang
data['utt'] = hstr
#
# main
#
rec.start_recording()
print("Please speak.")
while True:
samples = rec.get_samples()
audio, finalize = vad.process_audio(samples)
if not audio:
continue
logging.debug('decoding audio len=%d finalize=%s audio=%s' %
(len(audio), repr(finalize), audio[0].__class__))
user_utt, confidence = asr.decode(audio, finalize, stream_id=STREAM_ID)
# print("\r%s " % user_utt)
if finalize:
print("\r%s " % user_utt)
if findmathkeywords(user_utt):
print(calculate(user_utt))
if 'stop listening' in user_utt:
break
rec.start_recording()
finalize = False
recording = []
while not finalize:
samples = rec.get_samples()
audio, finalize = vad.process_audio(samples)
if not audio:
continue
recording.extend(audio)
user_utt, confidence = asr.decode(audio, finalize)
print "\r \rYou: %s " % user_utt,
if finalize and not user_utt:
finalize = False
recording = []
logging.info ("conv_user: %s" % user_utt)
rec.stop_recording()
print
# import pdb; pdb.set_trace()
ai_utt, score, action = kernal.process_input(ctx, user_utt)
chunks_length_ms = 30
chunks = make_chunks(audio_content, chunks_length_ms)
for samples in chunks:
sample = samples.get_array_of_samples()
sample = np.int16(sample)
if len(sample) < 480 :
add = np.zeros((480-len(sample),))
add=np.int16(add)
sample = np.concatenate((sample,add))
#audio, finalize = vad.process_audio(sample)
#if not audio:
# continue
logging.debug ('decoding audio len=%d finalize=%s audio=%s' % (len(sample), repr(finalize), sample[0].__class__))
user_utt, confidence = asr.decode(sample, finalize, stream_id=STREAM_ID)
#print "\r%s " % user_utt,
print "\r%s " % user_utt,
if finalize:
print "\r%s " % user_utt,
print "New Utterance\n"
vad = VAD(aggressiveness=aggressiveness)
print '\n \n \n'
print user_utt
rec.start_recording()
finalize = False
recording = []
while not finalize:
samples = rec.get_samples()
audio, finalize = vad.process_audio(samples)
if not audio:
continue
recording.extend(audio)
user_utt, confidence = asr.decode(audio, finalize)
print "\r \rYou: %s " % user_utt,
if finalize and not user_utt:
finalize = False
recording = []
logging.info("conv_user: %s" % user_utt)
rec.stop_recording()
print
# import pdb; pdb.set_trace()
ai_utt, score, action = kernal.process_input(ctx, user_utt)
finalize = False
recording = []
while not finalize:
samples = rec.get_samples()
audio, finalize = vad.process_audio(samples)
if not audio:
continue
recording.extend(audio)
user_utt, confidence = asr.decode(SAMPLE_RATE,
audio,
finalize,
stream_id=loc)
print "\r \rYou: %s " % user_utt,
if finalize and not user_utt:
finalize = False
recording = []
logging.info("conv_user: %s" % user_utt)
rec.stop_recording()
print
# import pdb; pdb.set_trace()
score, resps, actions, solutions, current_ctx = kernal.process_input(