基于树莓派的语音机器人

近年来语音识别发展迅速也带动了人工智能的发展。曾经渴望自己做一个机器人，但是无奈，心有余而力不足，经过多年的积累，小白的我也能用站着巨人的肩膀上玩下机器人了。

准备工作：树莓派，音频模块，stm32单片机，百度语音识别接口，喇叭。

整体思路：

1. 由于树莓派没有ADC模块，所以这里借助于stm32的ADC模块来实现将语音信号转换成数字信号，然后通过串口传 输 到树莓派你中，树莓派你将数据组装成wave文件，便于语音识别。

2. 通过http协议将组装的语音文件上传到百度语音识别平台进行识别。文档说明（免费调用）

3. 根据识别结果做出相应的处理。

4. 对于需要播放语音时，根据百度语音合成接口合成语音然后使用mplayer播放出来。mplayer安装参考 博客

部分代码：

将音频转换成wave文件

#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#include <wiringPi.h>
#include <wiringSerial.h>

#include <unistd.h>
#include <fcntl.h>
#include <sys/types.h>
#include <sys/ioctl.h>
#include "listen.h"
//gcc -o uart uart.c -lwiringPi
typedef struct WAV_Format WAVHEADER;

#define MAX_LISTEN_SIZES 1024*70 //定义接收数据的大小
#define bty 460800//串口的波特率
struct listen*listenUart()
{
 int fd,file;
 char buff,buff2;
 struct listen*liste=(struct listen*)malloc(sizeof(struct listen));

 unsigned short size;
 unsigned short*music,temp=0;
 unsigned short max=0,min=0;
 char*result=NULL;//存储最后的返回值
 int index=0,i=0;
 char stop=1;
 WAVHEADER wavHead;
 music=(unsigned short*)malloc(MAX_LISTEN_SIZES*2);
 result=(char*)malloc(MAX_LISTEN_SIZES*2+sizeof(WAVHEADER));
 if(wiringPiSetup() < 0)return NULL;
 if((fd = serialOpen ("/dev/ttyAMA0",bty))<0)
 {
 return NULL;
  printf("serial err\n");
 }
 //file=open("abc.wav", O_RDWR|O_CREAT);
 printf("oepn success\n");
 //serialPrintf(fd,"Hello World!!!");

 //需要对音频信号作出处理，当大于或者阈值时开始统计,知道录制完成
 int countTotal=0;
 int countNumber= 1000;//统计个数
 int countMax=2860;//最大值
 int countMin=2840;//最小值
 int startCount=1;
 while(1)
 {
  if(index==MAX_LISTEN_SIZES)
  {
  break;
  }
  buff=serialGetchar(fd);
  buff2=serialGetchar(fd);
  if((buff2&0x0F0)!=0)
  {
   buff2=serialGetchar(fd);
  }
  else
  {
   size=buff2;
   size=size<<8;
   size=(size&0xFF00)|(buff&0xFF);
   music[index]=size;
   if(startCount==1)
   {
   countTotal=countTotal+size;
   if(index>=countNumber)
   {
    int temp=countTotal/(countNumber+1);
    if(temp>countMax||temp<countMin)
    {
    startCount=0;
    //开始录音
    printf(":::::%d\n",temp);
    index++;
    }
    else
    {
    printf("temp:%d\n",temp);
    index=0;
    }
    countTotal=0;
   }
   else
   {
    index++;
   }
   }
   else
   {
    index++;
   }
  }
 }
 serialClose(fd);
 printf("end\n");
 //对音频进行放大处理
 max=music[0];
 min=music[0];
 for(i=i;i<MAX_LISTEN_SIZES;i++){
  temp=music[i];
  if(temp>max)
  {
  max=temp;
  }
  if(temp<min)
  {
  min=temp;
  }
 }

 size=max-min;

 for(i=0;i<MAX_LISTEN_SIZES;i++)
 {
  music[i]=(unsigned short)((music[i]-min)*1.0*6000/size);
 }
 wavHead.ChunkID=0x46464952; /* "RIFF" */
 wavHead.ChunkSize=sizeof(wavHead)+MAX_LISTEN_SIZES*2 -8; /* 36 + Subchunk2Size */
 wavHead.Format=0x45564157; /* "WAVE" */
 wavHead.Subchunk1ID=0x20746D66; /* "fmt " */
 wavHead.Subchunk1Size=0x10; /* 16 for PCM */
 wavHead.AudioFormat=0x01; /* PCM = 1*/
 wavHead.NumChannels=0x01; /* Mono = 1, Stereo = 2, etc. */
 wavHead.SampleRate=0x3E80; /* 8000, 44100, etc. */
 wavHead.ByteRate=0x7D00; /* = SampleRate * NumChannels * BitsPerSample/8 */
 wavHead.BlockAlign=0x02; /* = NumChannels * BitsPerSample/8 */
 wavHead.BitsPerSample=0x10; /* 8bits, 16bits, etc. */
 wavHead.Subchunk2ID=0x61746164; /* "data" */
 wavHead.Subchunk2Size=MAX_LISTEN_SIZES*2; /* data size */
 //返回数据赋值
 memcpy(result,(char*)&wavHead,sizeof(WAVHEADER));
 memcpy(result+sizeof(WAVHEADER),(char*)music,MAX_LISTEN_SIZES*2);
 liste->length=sizeof(WAVHEADER)+MAX_LISTEN_SIZES*2;
 liste->data=result;
 return liste;
}

将音频识别成文字

#include "convertText.h"
static Buffer *listen_buff2=NULL;
size_t listen_getData2(void *ptr, size_t size, size_t nmemb, void *stream)
{
 appendBuffer(listen_buff2,ptr,nmemb);
 return nmemb;
}
int listenText(char*result2)
{
 listen_buff2=initBuffer();
 struct listen*lsn=listenUart();
 char*base;
 int fileLength=lsn->length;
 int result=1;
 int baseSize=(lsn->length/3)*4+(lsn->length%3)*2+1;
 base=(char*)malloc(baseSize);
 base64_encode(lsn->data,lsn->length,base);
 //发送请求
 free(lsn->data);
 free(lsn);
 int code=initToken();
 if(code==1)
 {
 char*token=getToken();
 ///开始创建json字符串
 cJSON * root = cJSON_CreateObject();
 cJSON_AddItemToObject(root, "format", cJSON_CreateString("wav"));
 cJSON_AddItemToObject(root, "rate", cJSON_CreateString("16000"));
 cJSON_AddItemToObject(root, "channel", cJSON_CreateString("1"));
 cJSON_AddItemToObject(root, "cuid", cJSON_CreateString("34-68-95-91-77-43"));
 cJSON_AddItemToObject(root, "token", cJSON_CreateString(token));
 cJSON_AddItemToObject(root, "dev_pid", cJSON_CreateString("1537"));
 cJSON_AddItemToObject(root, "speech", cJSON_CreateString(base));
 cJSON_AddItemToObject(root, "len", cJSON_CreateNumber(fileLength));
 char*jsonParam=cJSON_PrintUnformatted(root);
  char*apiurl="http://vop.baidu.com/server_api";
  CURL* curl;
 CURLcode res;
  // ptr = curl_easy_escape(NULL, (char *)a, asize);
  curl = curl_easy_init();
 struct curl_slist* headers = NULL;
 headers = curl_slist_append(headers, "Content-Type:application/json");
 curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers);
 curl_easy_setopt(curl, CURLOPT_URL, apiurl);
 curl_easy_setopt(curl, CURLOPT_TIMEOUT, 60);
  curl_easy_setopt(curl, CURLOPT_POST, 1);
  //http://vop.baidu.com/server_api
  //CURLOPT_POSTFIELDS,CURLOPT_POSTFIELDSIZE
  curl_easy_setopt(curl, CURLOPT_POSTFIELDS, jsonParam);
  curl_easy_setopt(curl, CURLOPT_POSTFIELDSIZE, strlen(jsonParam));
 curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, listen_getData2);
 curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
 res = curl_easy_perform(curl);

 curl_easy_cleanup(curl);
 cJSON_Delete(root);
  curl_slist_free_all(headers);
 free(token);
 free(jsonParam);
 if (res == CURLE_OK)
  {
 char*chars;
 char*tempresult=(char*)malloc(listen_buff2->length+1);
 memcpy(tempresult,listen_buff2->buff,listen_buff2->length);
 tempresult[listen_buff2->length]=0;
 cJSON *json;
 cJSON * item = NULL;
 cJSON*errCode;
   json=cJSON_Parse(tempresult);
   item=cJSON_GetObjectItem(json, "result");
 errCode=cJSON_GetObjectItem(json, "err_no");
 if(errCode->valueint!=0)
 {
  return -3;
 }
 chars=cJSON_GetArrayItem(item,0)->valuestring;
 strcpy(result2,chars);
 free(tempresult);
 cJSON_Delete(json);
 return 0;
  }
 else
 {
 return -3;
 }

 }
 else
 {
 return -2;
 }
 return -1;
}

主程序

#include<stdio.h>
#include<string.h>
#include "convertText.h"
#include "mp3.h"
#include "led.h"
#include "say.h"

//gcc -o robot robot.o mp3.o Buffer.o base64.o token.o cJSON.o listen.o convertText.o led.o say.o -lcurl -lm -lwiringPi -lmad

void sayChina(char*china)
{
 int resp=initSay(china);
 printf("resp:%d\n",resp);
 if(resp==1)
 {
 int tte=playData("temp.mp3");
 printf("tte:%d\n",tte);
 }
}
int main()
{
 char text[100]={0};
 sayChina("你好，我是小志，有什么可以为你服务");
 while(1)
 {
 printf(";;;;;;;;");
 int code= listenText(text);
 if(code==0)
 {
 printf("result:%s\n",text);
 if(strstr(text,"播放音乐，")!=NULL||strstr(text,"打开音乐，")!=NULL)
 {
 sayChina("正在为你打开音乐");
 musicPlayFile("mu.mp3");
 }
 if(strstr(text,"打开灯，")!=NULL||strstr(text,"打开，")!=NULL)
 {
 sayChina("好的");
 printf("正在打开");
 ledOn();
 }
 if(strstr(text,"关闭灯，")!=NULL||strstr(text,"关闭，")!=NULL||strstr(text,"完毕，")!=NULL)
 {
 sayChina("好的");
 printf("正在关闭");
 ledOff();
 }
 if(strstr(text,"你叫什么")!=NULL||strstr(text,"你叫什么名字")!=NULL||strstr(text,"名字")!=NULL)
 {
 sayChina("我叫小志");
 }
 if(strstr(text,"今天天气咋样")!=NULL||strstr(text,"天气")!=NULL)
 {
 sayChina("外面在下雨，有点冷");
 }
 if(strstr(text,"中午好")!=NULL||strstr(text,"中午")!=NULL)
 {
 sayChina("好什么啊，我还没吃饭呢");
 }
 if(strstr(text,"你多大了")!=NULL||strstr(text,"今年几岁")!=NULL||strstr(text,"几岁")!=NULL)
 {
 sayChina("我才出生，还没满月");
 }
 }
 else
 {
 printf("error\n");
 }
 }
 return 0;
}

这里只是贴出来部分程序，所有代码请查看 链接 希望能和大家一起交流下心得。

以上就是本文的全部内容，希望对大家的学习有所帮助，也希望大家多多支持我们。