ros语音交互(五)移植科大讯飞语音识别到ros
将以前下载的的语音包的 samples/iat_record/的iat_record.c speech_recognizer.c speech_recognizer.c 拷贝到工程src中,
linuxrec.h speech_recognizer.h formats.h文件拷贝到 工程的include中
下面修改iat_record.c文件为xf_asr.cpp
/*
* xf_asr_node
* xf_asr.cpp
* 语音听写(iFly Auto Transform)技术能够实时地将语音转换成对应的文字。
*/ #include<ros/ros.h>
#include<std_msgs/String.h>
#include<std_msgs/Int32.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <unistd.h>
#include "qisr.h"
#include "msp_cmn.h"
#include "msp_errors.h"
#include "speech_recognizer.h" #define FRAME_LEN 640
#define BUFFER_SIZE 4096
#define ASRFLAG 1 using namespace std; bool flag = false;
bool recorder_Flag = true;
string result = ""; /* Upload User words */
static int upload_userwords()
{
char* userwords = NULL;
size_t len = 0;
size_t read_len = 0;
FILE* fp = NULL;
int ret = -1; fp = fopen("userwords.txt", "rb");
if (NULL == fp)
{
printf("\nopen [userwords.txt] failed! \n");
goto upload_exit;
} fseek(fp, 0, SEEK_END);
len = ftell(fp);
fseek(fp, 0, SEEK_SET); userwords = (char*)malloc(len + 1);
if (NULL == userwords)
{
printf("\nout of memory! \n");
goto upload_exit;
} read_len = fread((void*)userwords, 1, len, fp);
if (read_len != len)
{
printf("\nread [userwords.txt] failed!\n");
goto upload_exit;
}
userwords[len] = '\0'; MSPUploadData("userwords", userwords, len, "sub = uup, dtt = userword", &ret); //ÉÏ´«Óû§´Ê±í
if (MSP_SUCCESS != ret)
{
printf("\nMSPUploadData failed ! errorCode: %d \n", ret);
goto upload_exit;
} upload_exit:
if (NULL != fp)
{
fclose(fp);
fp = NULL;
}
if (NULL != userwords)
{
free(userwords);
userwords = NULL;
} return ret;
} static void show_result(char *str, char is_over)
{
printf("\rResult: [ %s ]", str);
if(is_over)
putchar('\n');
string s(str);
result = s;
flag = true; //设置发布话题为真
} static char *g_result = NULL;
static unsigned int g_buffersize = BUFFER_SIZE; void on_result(const char *result, char is_last)
{
if (result) {
size_t left = g_buffersize - 1 - strlen(g_result);
size_t size = strlen(result);
if (left < size) {
g_result = (char*)realloc(g_result, g_buffersize + BUFFER_SIZE);
if (g_result)
g_buffersize += BUFFER_SIZE;
else {
printf("mem alloc failed\n");
return;
}
}
strncat(g_result, result, size);
show_result(g_result, is_last);
}
}
void on_speech_begin()
{
if (g_result)
{
free(g_result);
}
g_result = (char*)malloc(BUFFER_SIZE);
g_buffersize = BUFFER_SIZE;
memset(g_result, 0, g_buffersize); printf("Start Listening...\n");
}
void on_speech_end(int reason)
{
if (reason == END_REASON_VAD_DETECT)
{
printf("\nSpeaking done \n");
recorder_Flag = false;
}
else
printf("\nRecognizer error %d\n", reason);
} /* demo recognize the audio from microphone */
static void demo_mic(const char* session_begin_params)
{
int errcode;
int i = 0; struct speech_rec iat; struct speech_rec_notifier recnotifier = {
on_result,
on_speech_begin,
on_speech_end
}; errcode = sr_init(&iat, session_begin_params, SR_MIC, &recnotifier);
if (errcode) {
printf("speech recognizer init failed\n");
return;
}
errcode = sr_start_listening(&iat);
if (errcode) {
printf("start listen failed %d\n", errcode);
}
/* demo 15 seconds recording */
while(recorder_Flag)
{
sleep(1);
}
errcode = sr_stop_listening(&iat);
if (errcode) {
printf("stop listening failed %d\n", errcode);
} sr_uninit(&iat);
} /*
* 打开麦克风 录音 发送到服务器
*/
void asrProcess()
{
int ret = MSP_SUCCESS;
int upload_on = 1; /* whether upload the user word */
/* login params, please do keep the appid correct */
const char* login_params = "appid = 57f49f64, work_dir = ."; /*
* See "iFlytek MSC Reference Manual"
*/
const char* session_begin_params =
"sub = iat, domain = iat, language = zh_cn, "
"accent = mandarin, sample_rate = 16000, "
"result_type = plain, result_encoding = utf8"; /* Login first. the 1st arg is username, the 2nd arg is password
* just set them as NULL. the 3rd arg is login paramertes
* */
ret = MSPLogin(NULL, NULL, login_params);
if (MSP_SUCCESS != ret) {
printf("MSPLogin failed , Error code %d.\n",ret);
goto exit; // login fail, exit the program
} /*
if (upload_on)
{
printf("Uploading the user words ...\n");
ret = upload_userwords();
if (MSP_SUCCESS != ret)
goto exit;
printf("Uploaded successfully\n");
}
*/ demo_mic(session_begin_params); exit:
MSPLogout(); // Logout...
} /*
* 根据发布的话题来修改录音标志
*/
void asrCallBack(const std_msgs::Int32::ConstPtr &msg)
{ ROS_INFO_STREAM("Topic is Subscriber");
if(msg->data == ASRFLAG)
{
asrProcess();
}
} /* main thread: start/stop record ; query the result of recgonization.
* record thread: record callback(data write)
* helper thread: ui(keystroke detection)
*/
int main(int argc, char* argv[])
{
ros::init(argc, argv, "xf_asr_node");
ros::NodeHandle nd; ros::Subscriber sub = nd.subscribe("/voice/xf_asr_topic", 1, asrCallBack);
ros::Publisher pub = nd.advertise<std_msgs::String>("/voice/tuling_arv_topic", 3); ros::Rate loop_rate(10); while(ros::ok())
{
if(flag)
{
std_msgs::String msg;
msg.data = result;
pub.publish(msg);
flag = false;
recorder_Flag = true;
} ros::spinOnce();
loop_rate.sleep();
} return 0;
}
Cmakefile 添加
add_executable(xf_asr_node src/xf_asr.cpp src/speech_recognizer.cpp src/linuxrec.cpp)
target_link_libraries(xf_asr_node ${catkin_LIBRARIES} -lmsc -lrt -ldl -lpthread -lasound)
编译后分别运行
$ rosrun tts_voice tts_voice_node
$ rosrun tts_voice tuling_arv_node
$ rosrun tts_voice xf_asr_node
$ rostopic pub -1 /voice/xf_asr_topic std_msgs/Int32 1

ros语音交互(五)移植科大讯飞语音识别到ros的更多相关文章
- ros语音交互(四)移植科大讯飞语音识别到ros
将以前下载的的语音包的 samples/iat_record/的iat_record.c speech_recognizer.c speech_recognizer.c 拷贝到工程src中, linu ...
- ROS语音交互(三)科大讯飞语音在ROS平台下使用
以上节tts语音输出为例 下载sdk链接:http://www.xfyun.cn/sdk/dispatcher 1.下载SDK,解压: 2.在ROS工作空间下创建一个Package: catkin_c ...
- ROS语音交互——科大讯飞语音合成TTS(二)
之前我用过科大讯飞的语音包,为了记录一下我重新使用一下 首先注册科大讯飞账号及应用,以后每个下载的在线使用SDK都是以此账户ID登录讯飞语音服务器. 下载科大讯飞在线合成包. $ unzip Linu ...
- ROS学习笔记五:创建和使用ROS msg和srv
1 msg和srv简介 1.1 msg文件 msg文件就是一个简单的text文件,其中每行有一个类型和名称,可用的类型如下: int8, int16, int32, int64 (plus uint* ...
- ROS语音交互(四)接入图灵语义理解
首先程序中会用到Json,curl 安装相应的库 $ sudo apt-get install libcurl3 libcurl4-openssl-dev$ sudo apt-get install ...
- ROS机器人语音交互(一)
语音交互早期已经广泛应用在手机端,电脑端,随着技术的成熟,接口逐渐开放,ROS上老外搞的开源语音识别只支持英文,识别率还低. 国内语音识别技术已经相当成熟稳定.感谢ros小课堂的讲解,解决了自己的疑惑 ...
- SLAM+语音机器人DIY系列:(七)语音交互与自然语言处理——1.语音交互相关技术
摘要 这一章将进入机器人语音交互的学习,让机器人能跟人进行语音对话交流.这是一件很酷的事情,本章将涉及到语音识别.语音合成.自然语言处理方面的知识.本章内容: 1.语音交互相关技术 2.机器人语音交互 ...
- SLAM+语音机器人DIY系列:(二)ROS入门——10.在实际机器人上运行ROS高级功能预览
摘要 ROS机器人操作系统在机器人应用领域很流行,依托代码开源和模块间协作等特性,给机器人开发者带来了很大的方便.我们的机器人“miiboo”中的大部分程序也采用ROS进行开发,所以本文就重点对ROS ...
- 安卓Android科大讯飞语音识别代码使用详解
科大讯飞的语音识别功能用在安卓代码中,我把语音识别写成了Service,然后在Fragment直接调用service服务.科大讯飞语音识别用的是带对话框的那个,直接调用科大讯飞的语音接口,代码采用链表 ...
随机推荐
- JS对象 window对象 屏幕可用高和宽度 1. screen.availWidth 属性返回访问者屏幕的宽度,以像素计,减去界面特性,比如任务栏。 2. screen.availHeight 属
屏幕可用高和宽度 1. screen.availWidth 属性返回访问者屏幕的宽度,以像素计,减去界面特性,比如任务栏. 2. screen.availHeight 属性返回访问者屏幕的高度,以像素 ...
- Navicat定时 同步数据库
转载:https://www.cnblogs.com/DFX339/p/11646812.html 我是用在了 mysql 和 sql server 数据同步(操作一模一样) 好文章,简单粗暴.适合我 ...
- 【学术篇】CF932E Team Work && bzoj5093 图的价值
两个题的传送门 对于CF这道题, 分别考虑每种可能的集合大小, 每个大小为\(k\)的集合数量有\(\binom nk\)个, 所以最后的答案就是 \[\sum_{i=0}^n\binom{n}{i} ...
- 转帖 新Eclipse安装与配置
Eclipse的官网地址:http://www.eclipse.org/ 我们下载J2EE版本:Eclipse IDE for Java EE Developers 目前最新版本是:Eclipse K ...
- vector注意事项
vector注意事项: 如果你添加元素,但是vector的容量(是容量,不是大小,容量比大小大,会预先多分配空间)不够的话,系统就会重新分配一段内存,然后把原先的内容全部拷贝过去,然后删除原先的内容. ...
- nodejs模块——目录操作
1.创建目录 使用fs.mkdir(path,[mode],callback)创建目录,path是需要创建的目录,[mode]是目录的权限(默认是0777),callback是回调函数. demo:m ...
- Django 自定义扩展命令
import datetime import logger from django.conf import settings from django.db.models import Q from d ...
- bootstrapTable 分页插件
前端: @{ ViewBag.Title = "BootstrapTable 入门"; Layout = null; } <!-- 引入bootstrap样式 --> ...
- python bs4解析网页时 bs4.FeatureNotFound: Couldn't find a tree builder with the features you requested: lxml. Do you need to inst(转)
Python小白,学习时候用到bs4解析网站,报错 bs4.FeatureNotFound: Couldn't find a tree builder with the features you re ...
- 2015 ACM/ICPC Asia Regional Changchun Online HDU - 5441 (离线+并查集)
题目:http://acm.hdu.edu.cn/showproblem.php?pid=5441 题意:给你n,m,k,代表n个城市,m条边,k次查询,每次查询输入一个x,然后让你一个城市对(u,v ...