百度AI接口的简单使用

百度云(https://cloud.baidu.com/)提供了许多人工智能的API接口,有一些是可以免费使用的。

image.png

环境配置

使用环境

windows7 x64
python3.7
vscode 编辑器，用于写代码
ffmpeg 用于将wav格式的音频文件转化成pcm格式，可在其官网(https://www.ffmpeg.org/)上下载
ffmpeg 下载后解压，不用安装，将bin中三个文件复制到system32目录下即可。
使用到的python包有
aip 百度AI接口库使用pip install baidu-aip安装
pyaudio python3.7使用pip安装会报错，可以下载whl文件(https://github.com/intxcc/pyaudio_portaudio/releases),离线安装pip install PyAudio-0.2.11-cp37-cp37m-win_amd64.whl，pyaudio库的使用可参考https://people.csail.mit.edu/hubert/pyaudio/，上面有例子代码
wave 使用pip install wave安装

代码实现

工具函数


#-*- coding:utf-8 -*-
"""音频处理"""
import os
import pyaudio
import wave
 
CHUNK = 1024
FORMAT = pyaudio.paInt16
CHANNELS = 2
#RATE = 44100
RATE = 16000
RECORD_SECONDS = 5
 
 
#用来播放声音，声音文件为wav格式
def play_wav(filename):
    wf = wave.open(filename, 'rb')
 
    p = pyaudio.PyAudio()
 
    stream = p.open(format=p.get_format_from_width(wf.getsampwidth()),
                    channels=wf.getnchannels(),
                    rate=wf.getframerate(),
                    output=True)
 
    data = wf.readframes(CHUNK)
 
 
    while len(data) != 0:
        stream.write(data)
        data = wf.readframes(CHUNK)
 
    stream.stop_stream()
    stream.close()
 
    p.terminate()
 
#录音，将声音保存为wav格式
def record(output_filename):
    p = pyaudio.PyAudio()
 
    stream = p.open(format=FORMAT,
                    channels=CHANNELS,
                    rate=RATE,
                    input=True,
                    frames_per_buffer=CHUNK)
 
    print("开始录音，最长为5s")
 
    frames = []
 
    over_flag = False
    for i in range(0, int(RATE / CHUNK * RECORD_SECONDS)):
        data = stream.read(CHUNK)
        frames.append(data)
 
    print("录音结束")
 
    stream.stop_stream()
    stream.close()
    p.terminate()
 
    wf = wave.open(output_filename, 'wb')
    wf.setnchannels(CHANNELS)
    wf.setsampwidth(p.get_sample_size(FORMAT))
    wf.setframerate(RATE)
    wf.writeframes(b''.join(frames))
    wf.close()
 
#调用ffmpeg将wav格式的音频文件转化成pcm格式，返回pcm格式文件名
#可参考https://www.cnblogs.com/DragonFire/p/9212935.html
#http://ai.baidu.com/docs#/ASR-Tool-convert/top
def wav2pcm(wav_file_name):
    # 假设 wav_file = "音频文件.wav"
    # wav_file.split(".") 得到["音频文件","wav"] 拿出第一个结果"音频文件"  与 ".pcm" 拼接 等到结果 "音频文件.pcm"
    pcm_file = "%s.pcm" %(wav_file_name.split(".")[0])
 
    # 就是此前我们在cmd窗口中输入命令,这里面就是在让Python帮我们在cmd中执行命令
    os.system("ffmpeg -y  -i %s  -acodec pcm_s16le -f s16le -ac 1 -ar 16000 %s > ffmpeg.log"%(wav_file_name,pcm_file))
 
    return pcm_file

测试代码


#-*- coding:utf-8 -*-
from aip import AipSpeech
import pyaudio
import wave
import sys
from pyaudio_utils import play_wav,record,wav2pcm
 
""" 你的 APPID AK SK """
#如何获取这三个参数可参考官方文档
APP_ID = 'xxxx'
API_KEY = 'xxxx'
SECRET_KEY = 'xxx'
 
client = AipSpeech(APP_ID, API_KEY, SECRET_KEY)
 
#语音合成
#参考https://cloud.baidu.com/doc/SPEECH/TTS-Online-Python-SDK.html#.E8.AF.B7.E6.B1.82.E8.AF.B4.E6.98.8E
def text2sound(msg):
    result  = client.synthesis(msg, 'zh', 1, {
        'vol': 5,'aue':6
    })
 
    # 识别正确返回语音二进制 错误则返回dict 参照下面错误码
    if not isinstance(result, dict):
        with open('auido.wav', 'wb') as f:
            f.write(result)
        play_wav('auido.wav')
 
#语音识别
#参考https://cloud.baidu.com/doc/SPEECH/ASR-Online-Python-SDK.html#.E8.AF.B7.E6.B1.82.E8.AF.B4.E6.98.8E
def sound2text(sound_filename,fmt):
    # 识别本地文件
    res = client.asr(open(sound_filename,'rb').read(), fmt, 16000, {
        'dev_pid': 1536,
    })
    print(res)
    if 'result' in res.keys():
        return(res['result'][0])
    return ''
 
 
 
def test():
    while True:
        #录音
        record('output.wav')
        #播放录音
        play_wav('output.wav')
        #将wav转化为pcm文件
        pcm_filename = wav2pcm('output.wav')
        #将声音转化为文字
        res = sound2text(pcm_filename,'pcm')
        if res:
            #将文字转化为声音并读出
            text2sound(res)
 
 
test()

参考资料

语音识别接口说明https://ai.baidu.com/ai-doc/SPEECH/Bk4o0bmt3
语音合成接口说明https://ai.baidu.com/ai-doc/SPEECH/Gk38y8lzk https://www.cnblogs.com/DragonFire/p/9212935.html https://blog.csdn.net/a506681571/article/details/85201279