class Main:
def __init__(self):
self.system_sound = SoundMP3('../sound/system_sound/registerd',
'../sound/system_sound/dev')
self.trans = Translate()
self.gc_controller = GoogleCloudController('', api_key='')
self.translater = Translate()
self.forecast = Forecast('')
self.mecab = MeCab.Tagger(
"-d /home/linuxbrew/.linuxbrew/lib/mecab/dic/mecab-ipadic-neologd")
self.temporary_data = ''
self.return_state = {}
self.jalan_controller = jalan_controller()
def analyze_text_by_juman(self, text):
res = subprocess.check_output('python3 JumanAnalyzer.py ' + text,
shell=True,
encoding='utf-8')
res = res.split('\n')
result_list = []
for item in res:
result_list.append(item.split(','))
return result_list[:-1]
def kanji2number(self, text):
text = text.replace('一', '1')
text = text.replace('二', '2')
text = text.replace('三', '3')
text = text.replace('四', '4')
text = text.replace('五', '5')
text = text.replace('六', '6')
text = text.replace('七', '7')
text = text.replace('八', '8')
text = text.replace('九', '9')
text = text.replace('十', '')
text = text.replace('百', '')
return text
def to_hiragana(self, text):
word1 = text
change_word1 = jaconv.kata2hira(word1)
return change_word1
def set_str2state(self, state):
state_list = state.split(' ')
for item in state_list:
item_list = item.split('=')
if len(item_list) == 2:
self.return_state[item_list[0]] = item_list[1]
def get_state(self):
result = ''
for key in self.return_state:
result += ' ' + str(key) + '=' + str(self.return_state[key])
return result
def analyze_text(self, text):
return self.mecab.parseToNode(text)
def pickup_where(self, node):
where = ''
is_first = True
while node:
word = node.surface
hinshi = node.feature.split(",")[2]
if hinshi == '地域':
if is_first:
where += word
is_first = False
else:
where += ' ' + word
node = node.next
return where
def analyze_whether_text(self, original_text):
node = self.analyze_text(original_text)
where = ''
when = ''
what = ''
where = self.pickup_where(node)
if '今日' in original_text:
when = '今日'
elif '明日' in original_text:
when = '明日'
elif '明後日' in original_text:
when = '明後日'
elif '曜日' in original_text:
index = original_text.index('曜日')
when = original_text[index - 1]
elif '今週' in original_text:
when = '今週'
if '天気' in original_text:
what = '天気'
if '詳細' in original_text:
what = '詳細'
return where, when, what
def match_start_str(self, validation_str):
start_li = [
'ええと', 'えっと', '絵本', 'エド', '五', 'エコ', 'Ｏ', 'でも', 'ゼロ', 'と。', 'Ｚ',
'それと', '是と', '図と'
]
for val in start_li:
if val in validation_str:
return True
return False
def sound_to_str(self, sound_file, mode):
return_text = ''
if mode == 'julius':
analyzer = VoiceAnalyzer(rate=16000)
return_text = analyzer.analyze_voice(sound_file)
elif mode == 'google':
result_text = self.gc_controller.speech2text(sound_file)
print(result_text)
if len(result_text) != 0:
return_text = result_text[0]
else:
return_text = ''
return return_text
def recording(self,
rate=16000,
threshhold=0.15,
record_seconds=1,
filtering=True,
mode='julius'):
sound_dir = '../sound/original/'
filtered_dir = '../sound/filtered/'
analyzer = VoiceAnalyzer(rate=rate,
threshold=threshhold,
record_seconds=record_seconds)
sound_file = analyzer.start_record(sound_dir) #開始させるボイス受付
if filtering:
treat_wave = Wave(sound_dir + sound_file,
filtered_dir + sound_file)
x = treat_wave.wave2numpy()
x = treat_wave.fourier_trans(x, 10000)
treat_wave.save_numpy2wave(x)
return self.sound_to_str(treat_wave.out_file, mode)
return self.sound_to_str(sound_dir + sound_file, mode)
def recognize_call(self, call_str='Z'):
while True:
rec_str = self.recording() #入力受付け
if self.match_start_str(rec_str):
return
else:
print(rec_str)
def get_where_foodname(self, rec_str):
text = self.analyze_text_by_juman(rec_str)
where = ''
food_name = ''
noun_result = []
noun_list = []
for item in text:
if item[4] == '地名':
where = item[0]
if item[3] == '名詞':
noun_list.append(item)
elif len(noun_list) != 0:
noun_result.append(noun_list)
noun_list = []
for result in noun_result:
is_food = False
for item in result:
if '食べ物' in item[7]:
is_food = True
break
if is_food:
if len(food_name) != 0:
food_name += ' '
for item in result:
food_name += item[0]
self.return_state['where'] = where
self.return_state['food_name'] = food_name
def classify_tasks(self, rec_str, user_state):
self.temporary_data = self.analyze_text(rec_str)
if '有名なもの' in rec_str or '有名な物' in rec_str:
node = self.temporary_data
where = self.pickup_where(node)
if len(where) != 0:
self.return_state['where'] = where
return Task.listen_drive_mode_content
elif '観光地' in rec_str:
node = self.temporary_data
where = self.pickup_where(node)
if len(where) != 0:
self.return_state['where'] = where
return Task.talk_kankou_list
else:
return Task.listen_kankou_where
elif '名物' in rec_str or '食べ物' in rec_str:
node = self.temporary_data
where = self.pickup_where(node)
if len(where) != 0:
self.return_state['where'] = where
return Task.talk_kankou_list
else:
return Task.listen_food_where
elif 'について教えて' in rec_str:
return Task.about_hearing
elif '行きたい' in rec_str or '案内して' in rec_str:
return Task.navigate
elif '食べたい' in rec_str:
self.get_where_foodname(rec_str)
return Task.talk_food_shop
elif '有名な' in rec_str:
return Task.talk_famous
if 'おはよう' in rec_str:
return Task.greeting
elif 'こんにちは' in rec_str:
return Task.greeting
elif 'こんばんは' in rec_str:
return Task.greeting
elif 'ストップ' in rec_str:
return Task.player_stop
elif '止めて' in rec_str:
return Task.player_stop
elif '消して' in rec_str:
return Task.player_stop
elif '閉じて' in rec_str:
return Task.player_stop
elif '天気' in rec_str:
return Task.weather
elif 'ニュース' in rec_str:
return Task.news
elif 'YouTube' in rec_str:
return Task.hear_youtube
return Task.default
def create_greeting_res(self, rec_str):
if 'おはよう' in rec_str:
return 'おはよう'
elif 'こんにちは' in rec_str:
return 'こんにちは'
elif 'こんばんは' in rec_str:
return 'こんばんは'
return 'ごきげんよう'
def create_weather_res(self, where, when, what):
return_text = ''
if when == '明日' or when == '今日' or when == '明後日' or when in [
'月', '火', '水', '木', '金', '土', '日'
]:
forecastDatetime = []
weatherDescription = []
temperature = []
dt_now = datetime.datetime.now()
forecast_day_diff = 0
if when == '今日':
pass
elif when == '明日':
forecast_day_diff = 1
elif when == '明後日':
forecast_day_diff = 2
else:
week_li = ['月', '火', '水', '木', '金', '土', '日']
weekday = datetime.date.today().weekday()
forecast_weekday = week_li.index(when)
forecast_day_diff = (forecast_weekday - weekday) % 7
dt_now += datetime.timedelta(days=forecast_day_diff)
data = self.translater.trans(where, 'ja', 'en')
if data != None:
data = data.text
else:
return 'すみません。翻訳時にエラーが起こりました。'
print(data, when, what)
forecastData = self.forecast.forecast_week(data)
if 'list' not in forecastData:
return 'where is invalid'
for item in forecastData['list']:
forecast_time = timezone('Asia/Tokyo').localize(
datetime.datetime.fromtimestamp(item['dt']))
if dt_now.day != forecast_time.day:
continue
else:
forecastDatetime.append(forecast_time)
weatherDescription.append(item['weather'][0]['main'])
temperature.append(item['main']['temp'])
print(forecastDatetime)
print(weatherDescription)
max_temp = int(max(temperature))
min_temp = int(min(temperature))
clear = 0
cloud = 0
lain = 0
morning = ''
afternoon = ''
for data in weatherDescription[:4]:
if data == 'Clear':
clear += 1
elif data == 'Clouds':
cloud += 1
else:
lain += 1
if clear > cloud and clear > lain:
morning = '晴れ'
elif cloud >= clear and cloud > lain:
morning = '曇り'
else:
morning = '雨'
clear = 0
cloud = 0
lain = 0
for data in weatherDescription[4:]:
if data == 'Clear':
clear += 1
elif data == 'Clouds':
cloud += 1
else:
lain += 1
if clear > cloud and clear > lain:
afternoon = '晴れ'
elif cloud >= clear and cloud > lain:
afternoon = '曇り'
else:
afternoon = '雨'
weather_text = ''
if morning == afternoon:
weather_text = morning
else:
weather_text = morning + 'のち' + afternoon
day = when
if when in ['月', '火', '水', '木', '金', '土', '日']:
day += '曜日'
return_text = '{0}の{1}の天気は{2}、最高気温{3}度、最低気温{4}度です。'.format(
day, where, weather_text, max_temp, min_temp)
return return_text
def create_news_res(self, voice_string):
news_string_index = voice_string.find('ニュース')
voice_string = voice_string[:news_string_index]
query = 'ニュース '
node = self.analyze_text(voice_string)
while node:
word = node.surface
hinshi = node.feature.split(",")[0]
if hinshi == '助詞':
pass
else:
query += word
node = node.next
url = self.gc_controller.get_youtube_url(query, order='relevance')
return url
def create_food_spot_res(self, task_name, voice_string):
res = ''
task = ''
if task_name == 'hear drive content':
if '食べ物' in voice_string or 'たべもの' in voice_string or '食い物' in voice_string or '名物' in voice_string:
where_analyze = self.analyze_text_by_juman(
self.return_state['where'])
parent_where = where_analyze[0][7].split(' ')[1].split(':')[2]
parent_where = parent_where.replace('都', '').replace(
'道', '').replace('府', '').replace('県', '')
where = where_analyze[0][0]
where, spot_name_list, spot_url_list = self.jalan_controller.get_specialty_food(
where, parent_where)
read_out_values = min(5, len(spot_name_list))
for i in range(len(spot_name_list)):
self.return_state['spot_name_' +
str(i)] = spot_name_list[i].replace(
' ', ' ')
self.return_state['spot_url_' + str(i)] = spot_url_list[i]
self.return_state['previous'] = 'read_food'
self.return_state['read_out_values_first'] = 0
self.return_state['read_out_values_last'] = read_out_values
res = where + 'で有名な食べ物を' + str(read_out_values) + '件読み上げます。'
for i in range(read_out_values):
res += str(i + 1) + '番、' + spot_name_list[i] + '。'
res += '詳細を聞きたい場合は、番号でおっしゃってください。'
task = 'hear number'
if len(spot_name_list) == 0:
res = '検索結果が0件でした。'
task = 'finish'
self.return_state['query_mode'] = 'food'
elif '観光地' in voice_string:
spot_name_list, spot_url_list = self.jalan_controller.query_search(
self.return_state['where'])
read_out_values = min(5, len(spot_name_list))
for i in range(len(spot_name_list)):
self.return_state['spot_name_' +
str(i)] = spot_name_list[i].replace(
' ', ' ')
self.return_state['spot_url_' + str(i)] = spot_url_list[i]
self.return_state['previous'] = 'read_kankou'
self.return_state['read_out_values_first'] = 0
self.return_state['read_out_values_last'] = read_out_values
res = 'スポットを' + str(read_out_values) + '件読み上げます。'
for i in range(read_out_values):
res += str(i + 1) + '番、' + spot_name_list[i] + '。'
res += '詳細を聞きたい場合は、番号でおっしゃってください。'
task = 'hear number'
if len(spot_name_list) == 0:
res = '検索結果が0件でした。'
task = 'finish'
return res, task
def create_where_detail_res(self, voice_string):
res = ''
task = ''
spot_name_list, spot_url_list = self.jalan_controller.query_search(
self.return_state['query'])
if len(spot_name_list) == 0:
res = 'すみません。わかりませんでした。'
task = 'finish'
else:
instruction, hours_text, where_text = self.jalan_controller.get_basic_info(
spot_url_list[0])
if instruction != None:
res = spot_name_list[0] + 'について読み上げます。' + instruction + '。'
if hours_text != '':
res += ' ' + hours_text
elif where_text != '':
res += '場所は、' + where_text
if len(res) == 0:
res = 'すみません。情報がありません。'
task = 'finish'
return res, task
def create_response(self, task_name, rec_str, lat, lng, user_state):
res = ''
task = ''
state = ''
if task_name == Task.weather:
where, when, what = self.analyze_whether_text(rec_str)
where = [where.replace('市', '')]
if len(where[0]) == 0 and lat != 0 and lng != 0:
where = []
data = self.gc_controller.reverse_geo_cording(lat, lng)
for d in data:
if 'locality' in d['types']:
where.append(d['short_name'])
if 'administrative_area_level_1' in d['types']:
where.append(d['short_name'])
if len(when) == 0:
res = 'いつの天気ですか？'
task = 'hear whether when'
state = 'where=' + str(where)
return res, task, state
elif len(where) == 0:
res = 'どこの天気ですか？'
task = 'hear whether where'
state = 'when=' + str(when)
return res, task, state
for data in where:
if len(data) == 0:
continue
res = self.create_weather_res(data, when, what)
if res == 'where is invalid':
res = 'すみません。場所がわかりませんでした。'
print(where, when, what)
elif task_name == Task.greeting:
res = self.create_greeting_res(rec_str)
elif task_name == Task.news:
state = 'youtube_url=' + self.create_news_res(rec_str).replace(
'=', '$equal')
print(state)
res = 'わかりました'
task = 'finish'
elif task_name == Task.hear_youtube:
task = 'hear youtube url'
res = '何を再生しますか？'
elif task_name == Task.player_stop:
task = 'stop player'
res = 'わかりました'
elif task_name == Task.listen_drive_mode_content:
where = self.return_state.get('where')
if where == None or where == '' and lat == 0 and lng == 0:
task = 'hear drive content where'
res = 'すみません。場所が取得できませんでした。どこの場所について聞きますか？'
else:
task = 'hear drive content'
res = '何に関して聞きますか？ たべもの、観光地に関してなら答えられます。'
elif task_name == Task.listen_kankou_where:
task = 'hear kankou where'
res = 'どこの観光地について聞きますか？'
elif task_name == Task.talk_kankou_list:
res, task = self.create_food_spot_res('hear drive content',
rec_str)
elif task_name == Task.about_hearing:
res = 'それは観光地ですか？ 食べ物ですか？'
task = 'hear about hearing mode'
self.return_state = {}
node = self.analyze_text(rec_str)
query = ''
while node:
word = node.surface
if word == 'について':
break
else:
hinshi = node.feature.split(",")[0]
if hinshi != '助詞':
query += word + ','
node = node.next
self.set_str2state('query=' + query)
elif task_name == Task.talk_food_shop:
where = self.return_state.get('where')
if self.return_state['food_name'] == '':
res = '食べ物の名前が取得できませんでした。'
task = 'finish'
elif where != None and where != '':
spot_name_list, spot_url_list = self.jalan_controller.query_search(
where + ' ' + self.return_state['food_name'], 'gourmet')
read_out_values = min(5, len(spot_name_list))
for i in range(len(spot_name_list)):
self.return_state['spot_name_' +
str(i)] = spot_name_list[i].replace(
' ', ' ')
self.return_state['spot_url_' + str(i)] = spot_url_list[i]
self.return_state['previous'] = 'read_kankou'
self.return_state['read_out_values_first'] = 0
self.return_state['read_out_values_last'] = read_out_values
res = 'お店を' + str(read_out_values) + '件読み上げます。'
for i in range(read_out_values):
res += str(i + 1) + '番、' + spot_name_list[i] + '。'
res += '詳細を聞きたい場合は、番号でおっしゃってください。'
task = 'hear number'
if len(spot_name_list) == 0:
res = '検索結果が0件でした。'
task = 'finish'
else:
res = 'どこについて聞きますか？'
task = 'listen food shop where'
elif task_name == Task.talk_famous:
self.get_where_foodname(rec_str)
where = self.return_state.get('where')
food_name = self.return_state.get('food_name')
if food_name != None and food_name != '':
if where != None and where != '':
spot_name_list, spot_url_list = self.jalan_controller.query_search(
where + ' ' + food_name, 'gourmet')
read_out_values = min(5, len(spot_name_list))
for i in range(len(spot_name_list)):
self.return_state['spot_name_' +
str(i)] = spot_name_list[i].replace(
' ', ' ')
self.return_state['spot_url_' +
str(i)] = spot_url_list[i]
self.return_state['previous'] = 'read_kankou'
self.return_state['read_out_values_first'] = 0
self.return_state['read_out_values_last'] = read_out_values
res = 'お店を' + str(read_out_values) + '件読み上げます。'
for i in range(read_out_values):
res += str(i + 1) + '番、' + spot_name_list[i] + '。'
res += '詳細を聞きたい場合は、番号でおっしゃってください。'
task = 'hear number'
if len(spot_name_list) == 0:
res = '検索結果が0件でした。'
task = 'finish'
else:
res = 'すみません。よくわかりません。'
task = 'finish'
else:
res = 'すみません。よくわかりません。'
task = 'finish'
elif task_name == Task.navigate:
res = 'すみません。この機能は開発中です。'
task = 'finish'
else:
res = 'すみません。よくわかりません。'
task = 'finish'
self.set_str2state(state)
return res, task
#return '明日の天気は雨のちくもり、最高気温15度、最低温度9度です。'
def sound_greeting(self, rec_str):
if 'おはよう' in rec_str:
self.system_sound.sound_registered_mp3('good_morning.mp3')
elif 'こんにちは' in rec_str:
self.system_sound.sound_registered_mp3('good_afternoon.mp3')
else:
self.system_sound.sound_registered_mp3('good_evening.mp3')
def return_result(self, voice_string, data, lat, lng):
res = ''
task = ''
self.set_str2state(data['state'])
removed_value = self.return_state.pop('youtube_url', None)
if 'でもない' in voice_string:
res = 'わかりました'
task = 'finish'
elif data['task'] == 'start_listen':
if self.match_start_str(voice_string):
res = 'ごようでしょうか'
task = 'normal hearing'
elif data['task'] == 'listen food shop where':
node = self.analyze_text(voice_string)
where = self.pickup_where(node)
print(where)
if where != None:
self.return_state['where'] = where
res, task = self.create_response(Task.talk_food_shop,
voice_string, lat, lng,
data['state'])
else:
res = 'すみません。場所がわかりませんでした。'
task = 'finish'
elif '他' in voice_string or 'それ以外' in voice_string or '次' in voice_string or 'もっと' in voice_string:
if self.return_state['previous'] == 'read_kankou':
spot_name_list = []
spot_url_list = []
dict_keys = self.return_state.keys()
i = 0
while (True):
if 'spot_name_' + str(i) in dict_keys:
spot_name_list.append(self.return_state['spot_name_' +
str(i)])
spot_url_list.append(self.return_state['spot_url_' +
str(i)])
i += 1
else:
break
read_out_values_first = int(
self.return_state['read_out_values_first'])
read_out_values_last = int(
self.return_state['read_out_values_last'])
if read_out_values_last == len(spot_name_list):
res = 'これ以上ありません。'
else:
next_last = min(read_out_values_last + 5,
len(spot_name_list))
res = '次の' + str(next_last -
read_out_values_last) + '件を読み上げます。'
for i in range(read_out_values_last, next_last):
res += str(i - read_out_values_last +
1) + '番、' + spot_name_list[i] + '。'
self.set_str2state('read_out_values_first=' +
str(read_out_values_last))
self.set_str2state('read_out_values_last=' +
str(next_last))
res += '詳細を聞きたい場合は、番号でおっしゃってください。'
task = 'hear number'
task = 'finish'
elif '番' in voice_string and ('聞きたい' in voice_string
or '教えて' in voice_string
or 'お願い' in voice_string):
if self.return_state.get('previous') == None:
res = 'すみません。履歴が残っていません。'
task = 'finish'
elif self.return_state['previous'] == 'read_kankou':
spot_name_list = []
spot_url_list = []
dict_keys = self.return_state.keys()
i = 0
while (True):
if 'spot_name_' + str(i) in dict_keys:
spot_name_list.append(self.return_state['spot_name_' +
str(i)])
spot_url_list.append(self.return_state['spot_url_' +
str(i)])
i += 1
else:
break
read_out_values_first = int(
self.return_state['read_out_values_first'])
read_out_values_last = int(
self.return_state['read_out_values_last'])
index = self.kanji2number(
voice_string[:voice_string.find('番')])
index = int(index) - 1 + read_out_values_first
if index >= len(spot_name_list):
res = 'その番号はありません。'
task = 'finish'
else:
self.set_str2state('detail_where_name=' +
spot_name_list[index])
self.set_str2state('detail_where_url=' +
spot_url_list[index])
instruction, hours_text, where_text = self.jalan_controller.get_basic_info(
spot_url_list[index])
if instruction != None:
res = spot_name_list[
index] + 'について読み上げます。' + instruction + '。'
if hours_text != '':
res += ' ' + hours_text
elif where_text != '':
res += '場所は、' + where_text
if len(res) == 0:
res = 'すみません。情報がありません。'
task = 'finish'
elif self.return_state['previous'] == 'read_food':
spot_name_list = []
spot_url_list = []
dict_keys = self.return_state.keys()
i = 0
while (True):
if 'spot_name_' + str(i) in dict_keys:
spot_name_list.append(self.return_state['spot_name_' +
str(i)])
spot_url_list.append(self.return_state['spot_url_' +
str(i)])
i += 1
else:
break
read_out_values_first = int(
self.return_state['read_out_values_first'])
read_out_values_last = int(
self.return_state['read_out_values_last'])
index = self.kanji2number(
voice_string[:voice_string.find('番')])
index = int(index) - 1 + read_out_values_first
if index >= len(spot_name_list):
res = 'その番号はありません。'
task = 'finish'
else:
self.set_str2state('detail_where_name=' +
spot_name_list[index])
self.set_str2state('detail_where_url=' +
spot_url_list[index])
instruction = self.jalan_controller.get_specialty_food_basic_info(
spot_url_list[index])
if instruction != None:
res = spot_name_list[
index] + 'について読み上げます。' + instruction + '。'
if len(res) == 0:
res = 'すみません。情報がありません。'
task = 'finish'
elif data['task'] == 'normal' or data['task'] == '':
task_name = self.classify_tasks(voice_string,
data['state']) #聞き取りに対する文章処理
res, task = self.create_response(task_name, voice_string, lat, lng,
data['state'])
elif data['task'] == 'hear kankou where':
node = self.analyze_text(voice_string)
where = self.pickup_where(node)
res, task = self.create_food_spot_res('hear drive content',
where + 'の観光地について教えて')
elif (data['task'] == 'hear whether where'):
when = data['state'][5:]
where = voice_string
if len(where) != 0:
where = where.replace('市', '')
what = ''
res = self.create_weather_res(where, when, what)
if res == 'where is invalid':
res = 'すみません。場所がわかりませんでした。'
task = 'finish'
elif (data['task'] == 'hear whether when'):
where = data['state'][6:]
if len(where) != 0:
where = where.replace('市', '')
when = voice_string
what = ''
res = self.create_weather_res(where, when, what)
if res == 'where is invalid':
res = 'すみません。場所がわかりませんでした。'
task = 'finish'
elif (data['task'] == 'hear youtube url'):
query = voice_string
url = self.gc_controller.get_youtube_url(
query, order='relevance').replace('=', '$equal')
self.set_str2state('youtube_url=' + url)
res = 'わかりました。'
task = 'finish'
elif data['task'] == 'hear drive content':
res, task = self.create_food_spot_res(data['task'], voice_string)
elif data['task'] == 'hear about hearing mode':
if '観光地' in voice_string:
res, task = self.create_where_detail_res(voice_string)
else:
res = 'すみません。場所がよくわかりませんでした。'
task = 'finish'
elif data['task'] == 'hear drive content where':
node = self.analyze_text(voice_string)
where = self.pickup_where(node)
if len(where) != 0:
self.return_state['where'] = where
task = 'hear drive content'
res = '何に関して聞きますか？ たべもの、観光地に関してなら答えられます。'
else:
task = 'finish'
res = 'すみません。場所がわかりませんでした。'
state = self.get_state()
print(res)
print(task)
print(state)
return res, task, state
def main_loop(self):
while (True):
self.recognize_call('Z') #起動聞き取り
print('success')
self.system_sound.sound_registered_mp3('start.mp3') #聞き取り音再生
time.sleep(0.1)
self.system_sound.sound_registered_mp3(
'start_record.mp3') #聞き取り開始音再生
time.sleep(0.5)
rec_str = self.recording(record_seconds=2,
filtering=False,
mode='google') #内容聞き取り
print(rec_str)
task_name = self.classify_tasks(rec_str) #聞き取りに対する文章処理
if task_name == Task.default:
self.system_sound.sound_registered_mp3('default.mp3')
elif task_name == Task.greeting:
self.sound_greeting(rec_str)
else:
res, task = self.create_response(task_name, rec_str) #返答生成
file_name = self.system_sound.create_sound(res)
self.system_sound.sound_dev_mp3(file_name)
time.sleep(0.5)
