python之調用科大訊飛的線上語音識别
沒注冊的話可以在官網注冊一下，新人注冊的話語音識别這塊有贈送的使用次數
科大訊飛
環境
cffi==1.12.3
gevent==1.4.0
greenlet==0.4.15
pycparser==2.19
six==1.12.0
websocket==0.2.1
websocket-client==0.56.0
wave
pyaudio
jieba
windows 直接
pip install xxx
或
pip3 isntall xxx
linux 的話 自行下載下傳pip 或 有python 3.6以及之後的配置一下即可.
代碼
關鍵詞
因為用到了
jieba
是以寫一個關鍵詞的類友善使用，可把關鍵詞讀取或寫入json檔案。
import os
import json
from typing import Dict
class KeyWord():
"""
keyval是一個數組
"""
def __init__(self,keyval=[],JSONfile=None,W_JSONfile=None):
self.keyval = keyval
self.JSONfile = JSONfile
self.W_JSONfile = W_JSONfile
def write(self):
dict_keyval ={"keyval":self.keyval}
json_dict = json.dumps(dict_keyval,ensure_ascii=False)
filePath = os.getcwd() + "/python/keywords.json"
if self.W_JSONfile is not None:
filePath = self.W_JSONfile
with open(filePath,'w') as fp:
fp.write(json_dict)
def read(self):
readF = os.getcwd() + "/python/keywords.json"
if self.JSONfile is not None:
readF = self.JSONfile
with open(readF,'r') as fp:
readstr = json.load(fp)
print("readstr = >",readstr,"\ntype =>",type(readstr))
list_str = list(readstr['keyval'])
return list_str
if __name__ == "__main__":
keyvals = ['碘伏','口罩','999感冒靈']
jsonstr = KeyWord(keyval=keyvals)
jsonstr.write()
keys = jsonstr.read()
# print("keys =>",keys)
for key in keys:
print(key)
錄音和格式轉化
因為語音識别嗎，得有語音檔案哈，是以寫了這個類友善錄音和格式轉化
#!/bin/python3
# 辨別引用的python版本
import pyaudio
import wave
import sys
import os
import numpy as np
"""
首先內建一下錄音功能和格式轉換功能
"""
class Record():
"""
錄音的類
CHUNK = 1024
FORMAT = pyaudio.paInt16
CHANNELS = 1 聲道
RATE = 16000 頻率
RECORD_SECONDS = 5 錄音時間 機關=> s
WAVE_OUTPUT_FILENAME = os.getcwd() + "/python/output1.wav" 錄音檔案
"""
def __init__(self,WAVE_OUTPUT_FILENAME,CHUNK=1024,
FORMAT=pyaudio.paInt16,CHANNELS=1,RECORD_SECONDS=5,
Input=True,RATE=16000,PCMName="out.pcm",DataType=np.int16):
self.CHUNK = CHUNK
self.FORMAT = FORMAT
self.CHANNELS = CHANNELS
self.RECORD_SECONDS = RECORD_SECONDS
self.WAVE_OUTPUT_FILENAME = WAVE_OUTPUT_FILENAME
self.Input = Input
self.RATE = RATE
self.PCMName = PCMName
self.DataType = DataType
def recording(self):
"""
這句代碼 會屏蔽一些不必要的報錯
os.close(sys.stderr.fileno())
"""
#隐藏一些報錯，這些不影響程式的運作
os.close(sys.stderr.fileno())
print("開始錄音")
p = pyaudio.PyAudio()
stream = p.open(format=self.FORMAT,
channels=self.CHANNELS,
rate=self.RATE,
input=self.Input,#預設為True
frames_per_buffer=self.CHUNK)
frames = []
for i in range(0, int(self.RATE / self.CHUNK * self.RECORD_SECONDS)):
data = stream.read(self.CHUNK)
frames.append(data)
print("done")
# 關閉流
stream.stop_stream()
stream.close()
p.terminate()
wf = wave.open(self.WAVE_OUTPUT_FILENAME, 'wb')
wf.setnchannels(self.CHANNELS)
wf.setsampwidth(p.get_sample_size(self.FORMAT))
wf.setframerate(self.RATE)
wf.writeframes(b''.join(frames))
wf.close()
def wav2pcm(self):
"""
音頻檔案wav格式 轉 pcm格式
"""
f = open(self.WAVE_OUTPUT_FILENAME, "rb")
f.seek(0)
f.read(1024)
data = np.fromfile(f, dtype=self.DataType)
# 擷取 分割後的 數組
filePath = str(self.WAVE_OUTPUT_FILENAME).split('/')
path = ''
# 拼接路徑 取出最後一位 [0,-1)
for item in filePath[:-1]:
path += item +'/'
path += self.PCMName
# print("PCM Path =>",path)
data.tofile(path)
print("結束")
# 可以傳回一個元組； 也可以把它封成數組傳回
return (self.WAVE_OUTPUT_FILENAME,path)
def run(self):
self.recording()
wavpath,path = self.wav2pcm()
# print("wave =>",wavpath,"\n","path =>",path)
# 這個就不寫入那個類裡了， 這樣友善調用 不需要再初始化類了
# 可直接copy到使用的類中或者檔案裡
def pcm2wav(pcmfile,wavfile,channels=1,rate=16000):
with open(pcmfile,'rb') as fp:
pcmdata = fp.read()
with wave.open(wavfile, 'wb') as wav:
wav.setnchannels(channels)
wav.setsampwidth(16 // 8)
wav.setframerate(rate)
# 寫入
wav.writeframes(pcmdata)
# 測試
if __name__ == "__main__":
wavepath = os.getcwd() + "/python/output1.wav"
dev = Record(wavepath)
# dev.run()
pcmfile = os.getcwd() + '/python/demo.pcm'
wavfile = os.getcwd() + '/python/demo.wav'
pcm2wav(pcmfile,wavfile)
科大訊飛線上語音識别
#!/bin/python3
"""
linux + Python3.7
"""
import datetime
import hashlib
import base64
import hmac
import json
from urllib.parse import urlencode
import time
import ssl
from wsgiref.handlers import format_date_time
from datetime import datetime
from time import mktime
import _thread as thread
import os
import websocket
import jieba as JB
from KeyWord import KeyWord
STATUS_FIRST_FRAME = 0 # 第一幀的辨別
STATUS_CONTINUE_FRAME = 1 # 中間幀辨別
STATUS_LAST_FRAME = 2 # 最後一幀的辨別
"""
為了友善以後複用，單獨內建 科大訊飛的 線上語音識别功能
"""
class OnlineASR():
"""
Automatic Speech Recognition (ASR)
語音識别
對官網執行個體的再封裝；
參數可到官網去擷取，這裡我寫成預設的了，可自行修改
"""
def __init__(self, AudioFile,
APPID='',
APISecret='',
APIKey=''):
"""
AudioFile = r''+os.getcwd()+'/python/output1.wav'
"""
self.JustTime = datetime.now()
self.AudioFile = AudioFile
self.APPID = APPID
self.APISecret = APISecret
self.APIKey = APIKey
# 公共參數(common)
self.CommonArgs = {"app_id": self.APPID}
# 業務參數(business)，更多個性化參數可在官網檢視
self.BusinessArgs = {"domain": "iat", "language": "zh_cn",
"accent": "mandarin", "vinfo": 1, "vad_eos": 10000}
#全局參數
self.RESULT = []
# jieba 初始化 詞庫
# 向分詞詞典增加新詞
KW = KeyWord()
keyvals = KW.read()
for key in keyvals:
JB.add_word(key)
# 運作類
def run(self):
wsUrl = self.create_url()
# print(wsUrl)
websocket.enableTrace(False)
ws = websocket.WebSocketApp(wsUrl,
on_message=self.on_message,
on_error=self.on_error,
on_close=self.on_close)
ws.on_open =self.on_open
ws.run_forever(sslopt={"cert_reqs": ssl.CERT_NONE})
CurrTime = datetime.now()
self.HandleEndStr(self.RESULT)
# print("lists =>",lists)
print("運作時間 => ", CurrTime-self.JustTime, "s")
return self.RESULT
# 生成url
def create_url(self):
url = 'wss://ws-api.xfyun.cn/v2/iat'
# 生成RFC1123格式的時間戳
now = datetime.now()
date = format_date_time(mktime(now.timetuple()))
# 拼接字元串
signature_origin = "host: " + "ws-api.xfyun.cn" + "\n"
signature_origin += "date: " + date + "\n"
signature_origin += "GET " + "/v2/iat " + "HTTP/1.1"
# 進行hmac-sha256進行加密
signature_sha = hmac.new(self.APISecret.encode('utf-8'), signature_origin.encode('utf-8'),
digestmod=hashlib.sha256).digest()
signature_sha = base64.b64encode(
signature_sha).decode(encoding='utf-8')
authorization_origin = "api_key=\"%s\", algorithm=\"%s\", headers=\"%s\", signature=\"%s\"" % (
self.APIKey, "hmac-sha256", "host date request-line", signature_sha)
authorization = base64.b64encode(
authorization_origin.encode('utf-8')).decode(encoding='utf-8')
# 将請求的鑒權參數組合為字典
v = {
"authorization": authorization,
"date": date,
"host": "ws-api.xfyun.cn"
}
# 拼接鑒權參數，生成url
url = url + '?' + urlencode(v)
# print("date: ",date)
# print("v: ",v)
# 此處列印出建立連接配接時候的url,參考本demo的時候可取消上方列印的注釋，比對相同參數時生成的url與自己代碼生成的url是否一緻
# print('websocket url :', url)
return url
# 處理傳回的字元串,把其加入數組中
def HandleMsg(self,msg):
# 編碼為json字元串 ensure_ascii 不是Ascii編碼
jsonstr = json.dumps(msg,ensure_ascii=False)
dictstr = json.loads(jsonstr)
res = ''
for item in dictstr:
# print("item => ",item)
parsestr = json.loads(json.dumps(item['cw'],ensure_ascii=False))[0]
end = json.loads(json.dumps(parsestr,ensure_ascii=False))
res += end['w']
self.RESULT.append(res)
# print("res =>",res)
# 處理最終的字元串; 使用jiaba
def HandleEndStr(self,msg):
# 因為隻是短短的一句話，故去掉多餘的字元
split_char = ''.join(msg)
chars = ['.','。',',','，',';','；',':','[',']','{','}',
'*','%','$','￥','@','^','……','-','=','+','!','~','`','<','>','《','》','?']
for item in chars:
split_char = str(split_char).replace(item,'')
# 處理完畢
# 使用 jieba 采用精确模式 ： 把文本精确的切分開，不存在備援單詞 ,傳回清單
msglist = JB.lcut(split_char)
self.RESULT = msglist
return msglist
# 收到websocket消息的處理
def on_message(self,ws, message):
try:
code = json.loads(message)["code"]
sid = json.loads(message)["sid"]
if code != 0:
errMsg = json.loads(message)["message"]
print("sid:%s call error:%s code is:%s" % (sid, errMsg, code))
else:
data = json.loads(message)["data"]["result"]["ws"]
# print(json.loads(message))
result = ""
for i in data:
for w in i["cw"]:
result += w["w"]
print("sid:%s call success!,data is:%s" % (sid, json.dumps(data, ensure_ascii=False)))
# json.dumps(data, ensure_ascii=False)
return self.HandleMsg(data)
except Exception as e:
print("receive msg,but parse exception:", e)
# 收到websocket錯誤的處理
def on_error(self,ws, error):
print("### error:", error)
# 收到websocket關閉的處理
def on_close(self,ws):
print("### closed ###")
# 收到websocket連接配接建立的處理
def on_open(self,ws):
def run(*args):
frameSize = 8000 # 每一幀的音頻大小
intervel = 0.04 # 發送音頻間隔(機關:s)
status = STATUS_FIRST_FRAME # 音頻的狀态資訊，辨別音頻是第一幀，還是中間幀、最後一幀
with open(self.AudioFile, "rb") as fp:
while True:
buf = fp.read(frameSize)
# 檔案結束
if not buf:
status = STATUS_LAST_FRAME
# 第一幀處理
# 發送第一幀音頻，帶business 參數
# appid 必須帶上，隻需第一幀發送
if status == STATUS_FIRST_FRAME:
d = {"common": self.CommonArgs,
"business": self.BusinessArgs,
"data": {"status": 0, "format": "audio/L16;rate=16000",
"audio": str(base64.b64encode(buf), 'utf-8'),
"encoding": "raw"}}
d = json.dumps(d)
ws.send(d)
status = STATUS_CONTINUE_FRAME
# 中間幀處理
elif status == STATUS_CONTINUE_FRAME:
d = {"data": {"status": 1, "format": "audio/L16;rate=16000",
"audio": str(base64.b64encode(buf), 'utf-8'),
"encoding": "raw"}}
ws.send(json.dumps(d))
# 最後一幀處理
elif status == STATUS_LAST_FRAME:
d = {"data": {"status": 2, "format": "audio/L16;rate=16000",
"audio": str(base64.b64encode(buf), 'utf-8'),
"encoding": "raw"}}
ws.send(json.dumps(d))
time.sleep(1)
break
# 模拟音頻采樣間隔
time.sleep(intervel)
ws.close()
thread.start_new_thread(run, ())
#測試
if __name__ == "__main__":
AudioFile = r''+os.getcwd()+'/python/output1.wav'
asr = OnlineASR(AudioFile=AudioFile,
APPID='',
APISecret='',
APIKey=''
)
res = asr.run()
print("test =>",res)
![TestLink導出用例轉換工具(XML2Excel)[圖]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)