├── README.md ├── asr.py ├── demo.py ├── getsound.py ├── gpt.py ├── resources ├── common.res ├── javs.pmdl ├── new.pmdl └── snowboy.umdl ├── snowboydecoder.py ├── sound ├── bye.wav ├── error.wav ├── hi.wav ├── result.wav ├── right.wav ├── start.wav └── wrong.wav └── tts.py /README.md: -------------------------------------------------------------------------------- 1 | # linux-gpt-assistant 2 | 欢迎学习交流 3 | -------------------------------------------------------------------------------- /asr.py: -------------------------------------------------------------------------------- 1 | # coding=utf-8 2 | 3 | import sys 4 | import json 5 | import base64 6 | import time 7 | import requests 8 | 9 | IS_PY3 = sys.version_info.major == 3 10 | 11 | 12 | from urllib.request import urlopen 13 | from urllib.request import Request 14 | from urllib.error import URLError 15 | from urllib.parse import urlencode 16 | timer = time.perf_counter 17 | 18 | #百度api 19 | API_KEY = '' 20 | SECRET_KEY = '' 21 | 22 | # 需要识别的文件 23 | AUDIO_FILE = '' # 只支持 pcm/wav/amr 格式，极速版额外支持m4a 格式 24 | # 文件格式 25 | FORMAT = "" # 文件后缀只支持 pcm/wav/amr 格式，极速版额外支持m4a 格式 26 | 27 | CUID = '123456PYTHON' 28 | # 采样率 29 | RATE = 16000 # 固定值 30 | 31 | # 普通版 32 | 33 | DEV_PID = 1537 # 1537 表示识别普通话，使用输入法模型。根据文档填写PID，选择语言及识别模型 34 | ASR_URL = 'http://vop.baidu.com/server_api' 35 | SCOPE = 'audio_voice_assistant_get' # 有此scope表示有asr能力，没有请在网页里勾选，非常旧的应用可能没有 36 | 37 | #测试自训练平台需要打开以下信息， 自训练平台模型上线后，您会看见 第二步：“”获取专属模型参数pid:8001，modelid:1234”，按照这个信息获取 dev_pid=8001，lm_id=1234 38 | # DEV_PID = 8001 ; 39 | # LM_ID = 1234 ; 40 | 41 | # 极速版 打开注释的话请填写自己申请的appkey appSecret ，并在网页中开通极速版（开通后可能会收费） 42 | 43 | # DEV_PID = 80001 44 | # ASR_URL = 'http://vop.baidu.com/pro_api' 45 | # SCOPE = 'brain_enhanced_asr' # 有此scope表示有极速版能力，没有请在网页里开通极速版 46 | 47 | # 忽略scope检查，非常旧的应用可能没有 48 | # SCOPE = False 49 | 50 | class DemoError(Exception): 51 | pass 52 | 53 | 54 | """ TOKEN start """ 55 | 56 | TOKEN_URL = 'http://aip.baidubce.com/oauth/2.0/token' 57 | 58 | 59 | def fetch_token(): 60 | params = {'grant_type': 'client_credentials', 61 | 'client_id': API_KEY, 62 | 'client_secret': SECRET_KEY} 63 | post_data = urlencode(params) 64 | if (IS_PY3): 65 | post_data = post_data.encode( 'utf-8') 66 | req = Request(TOKEN_URL, post_data) 67 | try: 68 | f = urlopen(req) 69 | result_str = f.read() 70 | except URLError as err: 71 | # print('token http response http code : ' + str(err.code)) 72 | result_str = err.read() 73 | if (IS_PY3): 74 | result_str = result_str.decode() 75 | 76 | # print(result_str) 77 | result = json.loads(result_str) 78 | # print(result) 79 | if ('access_token' in result.keys() and 'scope' in result.keys()): 80 | # print(SCOPE) 81 | if SCOPE and (not SCOPE in result['scope'].split(' ')): # SCOPE = False 忽略检查 82 | raise DemoError('scope is not correct') 83 | # print('SUCCESS WITH TOKEN: %s EXPIRES IN SECONDS: %s' % (result['access_token'], result['expires_in'])) 84 | return result['access_token'] 85 | else: 86 | raise DemoError('MAYBE API_KEY or SECRET_KEY not correct: access_token or scope not found in token response') 87 | 88 | """ TOKEN end """ 89 | 90 | def getasr(filename): 91 | AUDIO_FILE = filename 92 | FORMAT = AUDIO_FILE[-3:] 93 | 94 | token = fetch_token() 95 | speech_data = [] 96 | with open(AUDIO_FILE, 'rb') as speech_file: 97 | speech_data = speech_file.read() 98 | 99 | length = len(speech_data) 100 | if length == 0: 101 | raise DemoError('file %s length read 0 bytes' % AUDIO_FILE) 102 | speech = base64.b64encode(speech_data) 103 | speech = str(speech, 'utf-8') 104 | 105 | params = {'dev_pid': DEV_PID, 106 | #"lm_id" : LM_ID, #测试自训练平台开启此项 107 | 'format': FORMAT, 108 | 'rate': RATE, 109 | 'token': token, 110 | 'cuid': CUID, 111 | 'channel': 1, 112 | 'speech': speech, 113 | 'len': length 114 | } 115 | headers={ 116 | "Content-Type":'application/json' 117 | } 118 | r=requests.post(url=ASR_URL,headers=headers,data=json.dumps(params)) 119 | work=r.json() 120 | return(work['result'][0]) 121 | 122 | -------------------------------------------------------------------------------- /demo.py: -------------------------------------------------------------------------------- 1 | import snowboydecoder 2 | import sys 3 | import signal 4 | import getsound 5 | import os 6 | import time 7 | 8 | import tts 9 | import asr 10 | import gpt 11 | 12 | 13 | interrupted = False 14 | model =['resources/javs.pmdl','resources/new.pmdl'] 15 | 16 | os.close(sys.stderr.fileno()) 17 | 18 | callbacks = [lambda: callbacks1(), 19 | lambda: callbacks2()] 20 | 21 | def callbacks1(): 22 | msg=[{"role": "system", "content": "你说话的语气萌萌哒，用词十分可爱"}] 23 | detector.terminate() 24 | print("检测成功!") 25 | 26 | os.system('aplay -D default sound/hi.wav') 27 | try: 28 | getsound.rec() 29 | qu=asr.getasr('sound/question.wav') 30 | print('问:'+qu) 31 | msg.append({"role": "user", "content": qu}) 32 | if qu!='': 33 | an=gpt.getgpt(msg) 34 | print('答:'+an) 35 | tts.gettts(an) 36 | os.system('aplay -D default sound/result.wav') 37 | except: 38 | os.system('aplay -D default sound/error.wav') 39 | print("发生错误") 40 | 41 | os.system('aplay -D default sound/bye.wav') 42 | print("结束对话!") 43 | 44 | print('检测中...') 45 | detector.start(detected_callback=callbacks, 46 | interrupt_check=interrupt_callback, 47 | sleep_time=0.03) 48 | 49 | def callbacks2(): 50 | msg=[{"role": "system", "content": "你说话的语气萌萌哒，用词十分可爱"}] 51 | detector.terminate() 52 | print("检测成功!") 53 | 54 | os.system('aplay -D default sound/start.wav') 55 | try: 56 | getsound.rec() 57 | qu=asr.getasr('sound/question.wav') 58 | print('输入:'+qu) 59 | if '我爱你' in qu: 60 | os.system('aplay -D default sound/right.wav') 61 | for i in range(5): 62 | getsound.rec() 63 | qu=asr.getasr('sound/question.wav') 64 | print('问:'+qu) 65 | msg.append({"role": "user", "content": qu}) 66 | if '再见' in qu: break 67 | if qu=='':break 68 | an=gpt.getgpt(msg) 69 | msg.append({"role": "assistant", "content": an}) 70 | print('答:'+an) 71 | tts.gettts(an) 72 | os.system('aplay -D default sound/result.wav') 73 | else: 74 | os.system('aplay -D default sound/wrong.wav') 75 | print("密码错误") 76 | except: 77 | os.system('aplay -D default sound/error.wav') 78 | print("发生错误") 79 | 80 | os.system('aplay -D default sound/bye.wav') 81 | print("结束对话!") 82 | 83 | print('检测中...') 84 | detector.start(detected_callback=callbacks, 85 | interrupt_check=interrupt_callback, 86 | sleep_time=0.03) 87 | 88 | def signal_handler(signal, frame): 89 | global interrupted 90 | interrupted = True 91 | 92 | 93 | def interrupt_callback(): 94 | global interrupted 95 | return interrupted 96 | 97 | # capture SIGINT signal, e.g., Ctrl+C 98 | signal.signal(signal.SIGINT, signal_handler) 99 | 100 | detector = snowboydecoder.HotwordDetector(model, sensitivity=0.5) 101 | 102 | # main loop 103 | 104 | print('检测中...') 105 | detector.start(detected_callback=callbacks, 106 | interrupt_check=interrupt_callback, 107 | sleep_time=0.03) 108 | 109 | 110 | detector.terminate() -------------------------------------------------------------------------------- /getsound.py: -------------------------------------------------------------------------------- 1 | import speech_recognition as sr 2 | 3 | #从系统麦克风拾取音频数据，采样率为 16000 4 | def rec(): 5 | rate=16000 6 | r = sr.Recognizer() 7 | with sr.Microphone(sample_rate=rate) as source: 8 | print('正在获取声音中...') 9 | audio = r.listen(source,timeout=3,phrase_time_limit=13) 10 | with open("sound/question.wav", "wb") as f: 11 | f.write(audio.get_wav_data()) 12 | print('声音获取完成.') 13 | -------------------------------------------------------------------------------- /gpt.py: -------------------------------------------------------------------------------- 1 | import openai 2 | 3 | openai.api_key = "" # 你的 OpenAI API Key 4 | 5 | def getgpt(msg): 6 | completion = openai.ChatCompletion.create(model="gpt-3.5-turbo", 7 | messages=msg, 8 | max_tokens=2048, 9 | temperature=1.2) 10 | return completion.choices[0].message.content 11 | -------------------------------------------------------------------------------- /resources/common.res: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/VMIJUNV/linux-gpt-assistant/cd6761378de8e397183f8284324f02a5b579afe9/resources/common.res -------------------------------------------------------------------------------- /resources/javs.pmdl: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/VMIJUNV/linux-gpt-assistant/cd6761378de8e397183f8284324f02a5b579afe9/resources/javs.pmdl -------------------------------------------------------------------------------- /resources/new.pmdl: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/VMIJUNV/linux-gpt-assistant/cd6761378de8e397183f8284324f02a5b579afe9/resources/new.pmdl -------------------------------------------------------------------------------- /resources/snowboy.umdl: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/VMIJUNV/linux-gpt-assistant/cd6761378de8e397183f8284324f02a5b579afe9/resources/snowboy.umdl -------------------------------------------------------------------------------- /snowboydecoder.py: -------------------------------------------------------------------------------- 1 | #!/usr/bin/env python 2 | 3 | import collections 4 | import pyaudio 5 | import snowboydetect 6 | import time 7 | import wave 8 | import os 9 | import logging 10 | from ctypes import * 11 | from contextlib import contextmanager 12 | 13 | logging.basicConfig() 14 | logger = logging.getLogger("snowboy") 15 | logger.setLevel(logging.INFO) 16 | TOP_DIR = os.path.dirname(os.path.abspath(__file__)) 17 | 18 | RESOURCE_FILE = os.path.join(TOP_DIR, "resources/common.res") 19 | DETECT_DING = os.path.join(TOP_DIR, "resources/ding.wav") 20 | DETECT_DONG = os.path.join(TOP_DIR, "resources/dong.wav") 21 | 22 | def py_error_handler(filename, line, function, err, fmt): 23 | pass 24 | 25 | ERROR_HANDLER_FUNC = CFUNCTYPE(None, c_char_p, c_int, c_char_p, c_int, c_char_p) 26 | 27 | c_error_handler = ERROR_HANDLER_FUNC(py_error_handler) 28 | 29 | @contextmanager 30 | def no_alsa_error(): 31 | try: 32 | asound = cdll.LoadLibrary('libasound.so') 33 | asound.snd_lib_error_set_handler(c_error_handler) 34 | yield 35 | asound.snd_lib_error_set_handler(None) 36 | except: 37 | yield 38 | pass 39 | 40 | class RingBuffer(object): 41 | """Ring buffer to hold audio from PortAudio""" 42 | 43 | def __init__(self, size=4096): 44 | self._buf = collections.deque(maxlen=size) 45 | 46 | def extend(self, data): 47 | """Adds data to the end of buffer""" 48 | self._buf.extend(data) 49 | 50 | def get(self): 51 | """Retrieves data from the beginning of buffer and clears it""" 52 | tmp = bytes(bytearray(self._buf)) 53 | self._buf.clear() 54 | return tmp 55 | 56 | 57 | def play_audio_file(fname=DETECT_DING): 58 | """Simple callback function to play a wave file. By default it plays 59 | a Ding sound. 60 | 61 | :param str fname: wave file name 62 | :return: None 63 | """ 64 | ding_wav = wave.open(fname, 'rb') 65 | ding_data = ding_wav.readframes(ding_wav.getnframes()) 66 | with no_alsa_error(): 67 | audio = pyaudio.PyAudio() 68 | stream_out = audio.open( 69 | format=audio.get_format_from_width(ding_wav.getsampwidth()), 70 | channels=ding_wav.getnchannels(), 71 | rate=ding_wav.getframerate(), input=False, output=True) 72 | stream_out.start_stream() 73 | stream_out.write(ding_data) 74 | time.sleep(0.2) 75 | stream_out.stop_stream() 76 | stream_out.close() 77 | audio.terminate() 78 | 79 | 80 | class HotwordDetector(object): 81 | """ 82 | Snowboy decoder to detect whether a keyword specified by `decoder_model` 83 | exists in a microphone input stream. 84 | 85 | :param decoder_model: decoder model file path, a string or a list of strings 86 | :param resource: resource file path. 87 | :param sensitivity: decoder sensitivity, a float of a list of floats. 88 | The bigger the value, the more senstive the 89 | decoder. If an empty list is provided, then the 90 | default sensitivity in the model will be used. 91 | :param audio_gain: multiply input volume by this factor. 92 | :param apply_frontend: applies the frontend processing algorithm if True. 93 | """ 94 | 95 | def __init__(self, decoder_model, 96 | resource=RESOURCE_FILE, 97 | sensitivity=[], 98 | audio_gain=1, 99 | apply_frontend=False): 100 | 101 | tm = type(decoder_model) 102 | ts = type(sensitivity) 103 | if tm is not list: 104 | decoder_model = [decoder_model] 105 | if ts is not list: 106 | sensitivity = [sensitivity] 107 | model_str = ",".join(decoder_model) 108 | 109 | self.detector = snowboydetect.SnowboyDetect( 110 | resource_filename=resource.encode(), model_str=model_str.encode()) 111 | self.detector.SetAudioGain(audio_gain) 112 | self.detector.ApplyFrontend(apply_frontend) 113 | self.num_hotwords = self.detector.NumHotwords() 114 | 115 | if len(decoder_model) > 1 and len(sensitivity) == 1: 116 | sensitivity = sensitivity * self.num_hotwords 117 | if len(sensitivity) != 0: 118 | assert self.num_hotwords == len(sensitivity), \ 119 | "number of hotwords in decoder_model (%d) and sensitivity " \ 120 | "(%d) does not match" % (self.num_hotwords, len(sensitivity)) 121 | sensitivity_str = ",".join([str(t) for t in sensitivity]) 122 | if len(sensitivity) != 0: 123 | self.detector.SetSensitivity(sensitivity_str.encode()) 124 | 125 | self.ring_buffer = RingBuffer( 126 | self.detector.NumChannels() * self.detector.SampleRate() * 5) 127 | 128 | def start(self, detected_callback=play_audio_file, 129 | interrupt_check=lambda: False, 130 | sleep_time=0.03, 131 | audio_recorder_callback=None, 132 | silent_count_threshold=15, 133 | recording_timeout=100): 134 | """ 135 | Start the voice detector. For every `sleep_time` second it checks the 136 | audio buffer for triggering keywords. If detected, then call 137 | corresponding function in `detected_callback`, which can be a single 138 | function (single model) or a list of callback functions (multiple 139 | models). Every loop it also calls `interrupt_check` -- if it returns 140 | True, then breaks from the loop and return. 141 | 142 | :param detected_callback: a function or list of functions. The number of 143 | items must match the number of models in 144 | `decoder_model`. 145 | :param interrupt_check: a function that returns True if the main loop 146 | needs to stop. 147 | :param float sleep_time: how much time in second every loop waits. 148 | :param audio_recorder_callback: if specified, this will be called after 149 | a keyword has been spoken and after the 150 | phrase immediately after the keyword has 151 | been recorded. The function will be 152 | passed the name of the file where the 153 | phrase was recorded. 154 | :param silent_count_threshold: indicates how long silence must be heard 155 | to mark the end of a phrase that is 156 | being recorded. 157 | :param recording_timeout: limits the maximum length of a recording. 158 | :return: None 159 | """ 160 | self._running = True 161 | 162 | def audio_callback(in_data, frame_count, time_info, status): 163 | self.ring_buffer.extend(in_data) 164 | play_data = chr(0) * len(in_data) 165 | return play_data, pyaudio.paContinue 166 | 167 | with no_alsa_error(): 168 | self.audio = pyaudio.PyAudio() 169 | self.stream_in = self.audio.open( 170 | input=True, output=False, 171 | format=self.audio.get_format_from_width( 172 | self.detector.BitsPerSample() / 8), 173 | channels=self.detector.NumChannels(), 174 | rate=self.detector.SampleRate(), 175 | frames_per_buffer=2048, 176 | stream_callback=audio_callback) 177 | 178 | if interrupt_check(): 179 | logger.debug("detect voice return") 180 | return 181 | 182 | tc = type(detected_callback) 183 | if tc is not list: 184 | detected_callback = [detected_callback] 185 | if len(detected_callback) == 1 and self.num_hotwords > 1: 186 | detected_callback *= self.num_hotwords 187 | 188 | assert self.num_hotwords == len(detected_callback), \ 189 | "Error: hotwords in your models (%d) do not match the number of " \ 190 | "callbacks (%d)" % (self.num_hotwords, len(detected_callback)) 191 | 192 | logger.debug("detecting...") 193 | 194 | state = "PASSIVE" 195 | while self._running is True: 196 | if interrupt_check(): 197 | logger.debug("detect voice break") 198 | break 199 | data = self.ring_buffer.get() 200 | if len(data) == 0: 201 | time.sleep(sleep_time) 202 | continue 203 | 204 | status = self.detector.RunDetection(data) 205 | if status == -1: 206 | logger.warning("Error initializing streams or reading audio data") 207 | 208 | #small state machine to handle recording of phrase after keyword 209 | if state == "PASSIVE": 210 | if status > 0: #key word found 211 | self.recordedData = [] 212 | self.recordedData.append(data) 213 | silentCount = 0 214 | recordingCount = 0 215 | message = "Keyword " + str(status) + " detected at time: " 216 | message += time.strftime("%Y-%m-%d %H:%M:%S", 217 | time.localtime(time.time())) 218 | logger.info(message) 219 | callback = detected_callback[status-1] 220 | if callback is not None: 221 | callback() 222 | 223 | if audio_recorder_callback is not None: 224 | state = "ACTIVE" 225 | continue 226 | 227 | elif state == "ACTIVE": 228 | stopRecording = False 229 | if recordingCount > recording_timeout: 230 | stopRecording = True 231 | elif status == -2: #silence found 232 | if silentCount > silent_count_threshold: 233 | stopRecording = True 234 | else: 235 | silentCount = silentCount + 1 236 | elif status == 0: #voice found 237 | silentCount = 0 238 | 239 | if stopRecording == True: 240 | fname = self.saveMessage() 241 | audio_recorder_callback(fname) 242 | state = "PASSIVE" 243 | continue 244 | 245 | recordingCount = recordingCount + 1 246 | self.recordedData.append(data) 247 | 248 | logger.debug("finished.") 249 | 250 | def saveMessage(self): 251 | """ 252 | Save the message stored in self.recordedData to a timestamped file. 253 | """ 254 | filename = 'output' + str(int(time.time())) + '.wav' 255 | data = b''.join(self.recordedData) 256 | 257 | #use wave to save data 258 | wf = wave.open(filename, 'wb') 259 | wf.setnchannels(1) 260 | wf.setsampwidth(self.audio.get_sample_size( 261 | self.audio.get_format_from_width( 262 | self.detector.BitsPerSample() / 8))) 263 | wf.setframerate(self.detector.SampleRate()) 264 | wf.writeframes(data) 265 | wf.close() 266 | logger.debug("finished saving: " + filename) 267 | return filename 268 | 269 | def terminate(self): 270 | """ 271 | Terminate audio stream. Users can call start() again to detect. 272 | :return: None 273 | """ 274 | self.stream_in.stop_stream() 275 | self.stream_in.close() 276 | self.audio.terminate() 277 | self._running = False 278 | -------------------------------------------------------------------------------- /sound/bye.wav: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/VMIJUNV/linux-gpt-assistant/cd6761378de8e397183f8284324f02a5b579afe9/sound/bye.wav -------------------------------------------------------------------------------- /sound/error.wav: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/VMIJUNV/linux-gpt-assistant/cd6761378de8e397183f8284324f02a5b579afe9/sound/error.wav -------------------------------------------------------------------------------- /sound/hi.wav: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/VMIJUNV/linux-gpt-assistant/cd6761378de8e397183f8284324f02a5b579afe9/sound/hi.wav -------------------------------------------------------------------------------- /sound/result.wav: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/VMIJUNV/linux-gpt-assistant/cd6761378de8e397183f8284324f02a5b579afe9/sound/result.wav -------------------------------------------------------------------------------- /sound/right.wav: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/VMIJUNV/linux-gpt-assistant/cd6761378de8e397183f8284324f02a5b579afe9/sound/right.wav -------------------------------------------------------------------------------- /sound/start.wav: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/VMIJUNV/linux-gpt-assistant/cd6761378de8e397183f8284324f02a5b579afe9/sound/start.wav -------------------------------------------------------------------------------- /sound/wrong.wav: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/VMIJUNV/linux-gpt-assistant/cd6761378de8e397183f8284324f02a5b579afe9/sound/wrong.wav -------------------------------------------------------------------------------- /tts.py: -------------------------------------------------------------------------------- 1 | # coding=utf-8 2 | import sys 3 | import json 4 | 5 | IS_PY3 = sys.version_info.major == 3 6 | 7 | from urllib.request import urlopen 8 | from urllib.request import Request 9 | from urllib.error import URLError 10 | from urllib.parse import urlencode 11 | from urllib.parse import quote_plus 12 | 13 | #百度api 14 | API_KEY = '' 15 | SECRET_KEY = '' 16 | 17 | TEXT = "" 18 | 19 | # 发音人选择, 基础音库：0为度小美，1为度小宇，3为度逍遥，4为度丫丫， 20 | # 精品音库：5为度小娇，103为度米朵，106为度博文，110为度小童，111为度小萌，默认为度小美 21 | PER = 0 22 | # 语速，取值0-15，默认为5中语速 23 | SPD = 5 24 | # 音调，取值0-15，默认为5中语调 25 | PIT = 5 26 | # 音量，取值0-9，默认为5中音量 27 | VOL = 5 28 | # 下载的文件格式, 3：mp3(default) 4： pcm-16k 5： pcm-8k 6. wav 29 | AUE = 6 30 | 31 | FORMATS = {3: "mp3", 4: "pcm", 5: "pcm", 6: "wav"} 32 | FORMAT = FORMATS[AUE] 33 | 34 | CUID = "123456PYTHON" 35 | 36 | TTS_URL = 'http://tsn.baidu.com/text2audio' 37 | 38 | 39 | class DemoError(Exception): 40 | pass 41 | 42 | 43 | """ TOKEN start """ 44 | 45 | TOKEN_URL = 'http://aip.baidubce.com/oauth/2.0/token' 46 | SCOPE = 'audio_tts_post' # 有此scope表示有tts能力，没有请在网页里勾选 47 | 48 | 49 | def fetch_token(): 50 | # print("fetch token begin") 51 | params = {'grant_type': 'client_credentials', 52 | 'client_id': API_KEY, 53 | 'client_secret': SECRET_KEY} 54 | post_data = urlencode(params) 55 | if (IS_PY3): 56 | post_data = post_data.encode('utf-8') 57 | req = Request(TOKEN_URL, post_data) 58 | try: 59 | f = urlopen(req, timeout=5) 60 | result_str = f.read() 61 | except URLError as err: 62 | # print('token http response http code : ' + str(err.code)) 63 | result_str = err.read() 64 | if (IS_PY3): 65 | result_str = result_str.decode() 66 | 67 | # print(result_str) 68 | result = json.loads(result_str) 69 | # print(result) 70 | if ('access_token' in result.keys() and 'scope' in result.keys()): 71 | if not SCOPE in result['scope'].split(' '): 72 | raise DemoError('scope is not correct') 73 | # print('SUCCESS WITH TOKEN: %s ; EXPIRES IN SECONDS: %s' % (result['access_token'], result['expires_in'])) 74 | return result['access_token'] 75 | else: 76 | raise DemoError('MAYBE API_KEY or SECRET_KEY not correct: access_token or scope not found in token response') 77 | 78 | 79 | """ TOKEN end """ 80 | 81 | def gettts(text): 82 | TEXT=text 83 | token = fetch_token() 84 | tex = quote_plus(TEXT) # 此处TEXT需要两次urlencode 85 | # print(tex) 86 | params = {'tok': token, 'tex': tex, 'per': PER, 'spd': SPD, 'pit': PIT, 'vol': VOL, 'aue': AUE, 'cuid': CUID, 87 | 'lan': 'zh', 'ctp': 1} # lan ctp 固定参数 88 | 89 | data = urlencode(params) 90 | # print('test on Web Browser' + TTS_URL + '?' + data) 91 | 92 | req = Request(TTS_URL, data.encode('utf-8')) 93 | has_error = False 94 | try: 95 | f = urlopen(req) 96 | result_str = f.read() 97 | 98 | headers = dict((name.lower(), value) for name, value in f.headers.items()) 99 | 100 | has_error = ('content-type' not in headers.keys() or headers['content-type'].find('audio/') < 0) 101 | except URLError as err: 102 | # print('asr http response http code : ' + str(err.code)) 103 | result_str = err.read() 104 | has_error = True 105 | 106 | with open('sound/result.wav', 'wb') as of: 107 | of.write(result_str) 108 | 109 | --------------------------------------------------------------------------------