(百度的rest接口的部分网址发生了一定的变化,相关代码已更新)
百度通过 REST API 的方式给开发者提供一个通用的 HTTP 接口,基于该接口,开发者可以轻松的获得语音合成与语音识别能力。SDK中只提供了PHP、C和JAVA的相关样例,使用python也可以灵活的对端口进行调用,本文描述了简单使用Python调用百度语音识别服务 REST API 的简单样例。
1、语音识别与语音合成的调用
注册开发者帐号和创建应用的过程就不再赘述,百度的REST API在调用过程基本分为三步:
- 获取token
- 向Rest接口提交数据
- 处理返回数据
具体代码如下所示:
#!/usr/bin/python3
import urllib.request
import urllib
import json
import base64
class BaiduRest:
def __init__(self, cu_id, api_key, api_secert):
# token认证的url
self.token_url = "https://openapi.baidu.com/oauth/2.0/token?grant_type=client_credentials&client_id=%s&client_secret=%s"
# 语音合成的resturl
self.getvoice_url = "http://tsn.baidu.com/text2audio?tex=%s&lan=zh&cuid=%s&ctp=1&tok=%s"
# 语音识别的resturl
self.upvoice_url = 'http://vop.baidu.com/server_api'
self.cu_id = cu_id
self.getToken(api_key, api_secert)
return
def getToken(self, api_key, api_secert):
# 1.获取token
token_url = self.token_url % (api_key,api_secert)
r_str = urllib.request.urlopen(token_url).read()
token_data = json.loads(r_str)
self.token_str = token_data['access_token']
pass
def getVoice(self, text, filename):
# 2. 向Rest接口提交数据
get_url = self.getvoice_url % (urllib.parse.quote(text), self.cu_id, self.token_str)
voice_data = urllib.request.urlopen(get_url).read()
# 3.处理返回数据
voice_fp = open(filename,'wb+')
voice_fp.write(voice_data)
voice_fp.close()
pass
def getText(self, filename):
# 2. 向Rest接口提交数据
data = {}
# 语音的一些参数
data['format'] = 'wav'
data['rate'] = 8000
data['channel'] = 1
data['cuid'] = self.cu_id
data['token'] = self.token_str
wav_fp = open(filename,'rb')
voice_data = wav_fp.read()
data['len'] = len(voice_data)
data['speech'] = base64.b64encode(voice_data).decode('utf-8')
post_data = json.dumps(data)
r_data = urllib.request.urlopen(self.upvoice_url,data=bytes(post_data,encoding="utf-8")).read()
# 3.处理返回数据
return json.loads(r_data)['result']
if __name__ == "__main__":
# 我的api_key,供大家测试用,在实际工程中请换成自己申请的应用的key和secert
api_key = "SrhYKqzl3SE1URnAEuZ0FKdT"
api_secert = "hGqeCkaMPb0ELMqtRGc2VjWdmjo7T89d"
# 初始化
bdr = BaiduRest("test_python", api_key, api_secert)
# 将字符串语音合成并保存为out.mp3
bdr.getVoice("你好北京邮电大学!", "out.mp3")
# 识别test.wav语音内容并显示
print(bdr.getText("out.wav"))
2、调用pyaudio使用麦克风录制声音
python中的pyaudio库可以直接通过麦克风录制声音,可使用pip进行安装。我们可以通过调用该库,获取到wav测试语音。
具体代码如下所示:
#!/usr/bin/python3
# -*- coding: utf-8 -*-
from pyaudio import PyAudio, paInt16
import numpy as np
from datetime import datetime
import wave
class recoder:
NUM_SAMPLES = 2000 #pyaudio内置缓冲大小
SAMPLING_RATE = 8000 #取样频率
LEVEL = 500 #声音保存的阈值
COUNT_NUM = 20 #NUM_SAMPLES个取样之内出现COUNT_NUM个大于LEVEL的取样则记录声音
SAVE_LENGTH = 8 #声音记录的最小长度:SAVE_LENGTH * NUM_SAMPLES 个取样
TIME_COUNT = 60 #录音时间,单位s
Voice_String = []
def savewav(self,filename):
wf = wave.open(filename, 'wb')
wf.setnchannels(1)
wf.setsampwidth(2)
wf.setframerate(self.SAMPLING_RATE)
wf.writeframes(np.array(self.Voice_String).tostring())
# wf.writeframes(self.Voice_String.decode())
wf.close()
def recoder(self):
pa = PyAudio()
stream = pa.open(format=paInt16, channels=1, rate=self.SAMPLING_RATE, input=True,
frames_per_buffer=self.NUM_SAMPLES)
save_count = 0
save_buffer = []
time_count = self.TIME_COUNT
while True:
time_count -= 1
# print time_count
# 读入NUM_SAMPLES个取样
string_audio_data = stream.read(self.NUM_SAMPLES)
# 将读入的数据转换为数组
audio_data = np.fromstring(string_audio_data, dtype=np.short)
# 计算大于LEVEL的取样的个数
large_sample_count = np.sum( audio_data > self.LEVEL )
print(np.max(audio_data))
# 如果个数大于COUNT_NUM,则至少保存SAVE_LENGTH个块
if large_sample_count > self.COUNT_NUM:
save_count = self.SAVE_LENGTH
else:
save_count -= 1
if save_count < 0:
save_count = 0
if save_count > 0 :
# 将要保存的数据存放到save_buffer中
#print save_count > 0 and time_count >0
save_buffer.append( string_audio_data )
else:
#print save_buffer
# 将save_buffer中的数据写入WAV文件,WAV文件的文件名是保存的时刻
#print "debug"
if len(save_buffer) > 0 :
self.Voice_String = save_buffer
save_buffer = []
print("Recode a piece of voice successfully!")
return True
if time_count==0:
if len(save_buffer)>0:
self.Voice_String = save_buffer
save_buffer = []
print("Recode a piece of voice successfully!")
return True
else:
return False
if __name__ == "__main__":
r = recoder()
r.recoder()
r.savewav("test.wav")