ROS科大讯飞语音（二）说话篇

最新推荐文章于 2024-07-12 13:14:50 发布

大吉机器人

最新推荐文章于 2024-07-12 13:14:50 发布

阅读量739

点赞数 1

ROS科大讯飞语音（二）说话篇

https://blog.csdn.net/zhouge94/article/details/52077997

基本环境配置

创建工作空间

$ mkdir -p ~/catkin_ws/src
$ cd ~/catkin_ws/src
$ catkin_init_workspace

首次编译

$ cd ~/catkin_ws/
$ catkin_make

创建语音包

$ cd src/
$ catkin_create_pkg xf_voice std_msgs rospy roscpp

安装语音库

创建/Robot文件夹将所有者改为当前用户

$ sudo mkdir /Robot
$ sudo chown zhouge /Robot/

然后将你下载的语音sdk中的libmsc.so放到/Robot/voice/lib/文件夹下
将bin目录和lib/inc目录都放到/Robot/voice/文件夹下，
安装mpalyer播放器
创建/Robot/voice/wav和/Robot/cmd文件夹，前者用来保存临时音频文件，后者用来保存管道通信文件。

$ sudo apt-get install mplayer
$ mkdir /Robot/voice/wav
$ mkdir /Robot/cmd

至此你的/Robot文件夹应该有cmd和voice两个文件夹，voice下应该有bin inc lib wav四个文件夹
然后我们把语音库放到系统库文件夹/usr/lib/

$ sudo cp /Robot/voice/lib/libmsc.so /usr/lib/

编写语音合成节点

源码

回到工作空间目录，~/catkin_ws
在~/catkin_ws/src/xf_voice/src目录下新建文件xf_tts.cpp 并将以下内容复制进去
注意：将里面的appid改成你自己的；

#include <stdio.h>
#include <unistd.h>
#include <errno.h>
#include "/Robot/voice/inc/qtts.h"
#include "/Robot/voice/inc/msp_cmn.h"
#include "/Robot/voice/inc/msp_errors.h"
#include "ros/ros.h"
#include "std_msgs/String.h"
#include <sstream>
#include <sys/types.h>
#include <sys/stat.h>
#define SAYIT  system("cp  /Robot/voice/wav/say.wav /Robot/voice/wav/temp.wav>/Robot/cmd/Mplayer_cmd");system("echo loadfile /Robot/voice/wav/temp.wav>/Robot/cmd/Mplayer_cmd")
typedef int SR_DWORD;
typedef short int SR_WORD ;
/* wav音频头部格式 */
typedef struct _wave_pcm_hdr
{
    char            riff[4];                // = "RIFF"
    int             size_8;                 // = FileSize - 8
    char            wave[4];                // = "WAVE"
    char            fmt[4];                 // = "fmt "
    int             fmt_size;               // = 下一个结构体的大小 : 16

    short int       format_tag;             // = PCM : 1
    short int       channels;               // = 通道数 : 1
    int             samples_per_sec;        // = 采样率 : 8000 | 6000 | 11025 | 16000
    int             avg_bytes_per_sec;      // = 每秒字节数 : samples_per_sec * bits_per_sample / 8
    short int       block_align;            // = 每采样点字节数 : wBitsPerSample / 8
    short int       bits_per_sample;        // = 量化比特数: 8 | 16

    char            data[4];                // = "data";
    int             data_size;              // = 纯数据长度 : FileSize - 44 
} wave_pcm_hdr;
/* 默认wav音频头部数据 */
wave_pcm_hdr default_wav_hdr = 
{
    { 'R', 'I', 'F', 'F' },
    0,
    {'W', 'A', 'V', 'E'},
    {'f', 'm', 't', ' '},
    16,
    1,
    1,
    16000,
    32000,
    2,
    16,
    {'d', 'a', 't', 'a'},
    0  
};
/* 文本合成 */
int text_to_speech(const char* src_text, const char* des_path, const char* params)
{
    int          ret          = -1;
    FILE*        fp           = NULL;
    const char*  sessionID    = NULL;
    unsigned int audio_len    = 0;
    wave_pcm_hdr wav_hdr      = default_wav_hdr;
    int          synth_status = MSP_TTS_FLAG_STILL_HAVE_DATA;

    if (NULL == src_text || NULL == des_path)
    {
        printf("params is error!\n");
        return ret;
    }
    fp = fopen(des_path, "wb");
    if (NULL == fp)
    {
        printf("open %s error.\n", des_path);
        return ret;
    }
    /* 开始合成 */
    sessionID = QTTSSessionBegin(params, &ret);
    if (MSP_SUCCESS != ret)
    {
        printf("QTTSSessionBegin failed, error code: %d.\n", ret);
        fclose(fp);
        return ret;
    }
    ret = QTTSTextPut(sessionID, src_text, (unsigned int)strlen(src_text), NULL);
    if (MSP_SUCCESS != ret)
    {
        printf("QTTSTextPut failed, error code: %d.\n",ret);
        QTTSSessionEnd(sessionID, "TextPutError");
        fclose(fp);
        return ret;
    }
    printf("正在合成 ...\n");
    fwrite(&wav_hdr, sizeof(wav_hdr) ,1, fp); //添加wav音频头，使用采样率为16000
    while (1) 
    {
        /* 获取合成音频 */
        const void* data = QTTSAudioGet(sessionID, &audio_len, &synth_status, &ret);
        if (MSP_SUCCESS != ret)
            break;
        if (NULL != data)
        {
            fwrite(data, audio_len, 1, fp);
            wav_hdr.data_size += audio_len; //计算data_size大小
        }
        if (MSP_TTS_FLAG_DATA_END == synth_status)
            break;
    }//合成状态synth_status取值请参阅《讯飞语音云API文档》
    printf("\n");
    if (MSP_SUCCESS != ret)
    {
        printf("QTTSAudioGet failed, error code: %d.\n",ret);
        QTTSSessionEnd(sessionID, "AudioGetError");
        fclose(fp);
        return ret;
    }
    /* 修正wav文件头数据的大小 */
    wav_hdr.size_8 += wav_hdr.data_size + (sizeof(wav_hdr) - 8);
    /* 将修正过的数据写回文件头部,音频文件为wav格式 */
    fseek(fp, 4, 0);
    fwrite(&wav_hdr.size_8,sizeof(wav_hdr.size_8), 1, fp); //写入size_8的值
    fseek(fp, 40, 0); //将文件指针偏移到存储data_size值的位置
    fwrite(&wav_hdr.data_size,sizeof(wav_hdr.data_size), 1, fp); //写入data_size的值
    fclose(fp);
    fp = NULL;
    /* 合成完毕 */
    ret = QTTSSessionEnd(sessionID, "Normal");
    if (MSP_SUCCESS != ret)
    {
        printf("QTTSSessionEnd failed, error code: %d.\n",ret);
    }
    return ret;
}

int xf_tts(const char* text,const char *filename)
{
    int         ret                  = MSP_SUCCESS;
    const char* login_params         = "appid = 573bdbff, work_dir = .";//登录参数,appid与msc库绑定,请勿随意改动
    const char* session_begin_params = "engine_type =local, text_encoding = UTF8, tts_res_path = fo|/Robot/voice/bin/msc/res/tts/xiaoyan.jet;fo|/Robot/voice/bin/msc/res/tts/common.jet, sample_rate = 16000, speed = 50, volume = 50, pitch = 50, rdn = 2";
    /* 用户登录 */
    ret = MSPLogin(NULL, NULL, login_params); //第一个参数是用户名，第二个参数是密码，第三个参数是登录参数，用户名和密码可在http://open.voicecloud.cn注册获取
    if (MSP_SUCCESS != ret)
    {
        printf("MSPLogin failed, error code: %d.\n", ret);
        goto exit ;//登录失败，退出登录
    }
    /* 文本合成 */
    printf("开始合成 ...\n");
    ret = text_to_speech(text, filename, session_begin_params);
    if (MSP_SUCCESS != ret)
    {
        printf("text_to_speech failed, error code: %d.\n", ret);
    }
    printf("合成完毕\n");

exit:
    MSPLogout(); //退出登录
    return 0;
}
void xfcallback(const std_msgs::String::ConstPtr& msg)
{
  char cmd[2000];
  std::cout<<"I heard,I will say:"<<msg->data.c_str()<<std::endl;
  xf_tts(msg->data.c_str(),"/Robot/voice/wav/say.wav");
  sprintf(cmd,"echo %s>/Robot/cmd/saywords",msg->data.c_str());
  popen(cmd,"r");
  SAYIT;
}
int main(int argc,char **argv)
{
    unlink("/Robot/cmd/Mplayer_cmd");
    mkfifo("/Robot/cmd/Mplayer_cmd", 0777);
    popen("mplayer -quiet -slave -input file=/Robot/cmd/Mplayer_cmd -idle","r");
    printf("Mplayer Run Success");
    const char* filename        = "/Robot/voice/wav/say.wav"; //合成的语音文件名称
    const char* text                 = "语音合成模块启动成功！"; //合成文本
    xf_tts(text,filename);
    SAYIT;
    ros::init(argc,argv,"xf_tts");
    ros::NodeHandle n;
    ros::Subscriber sub =n.subscribe("xfsaywords",1000,xfcallback);
    ros::spin();
    return 0;
}

CMakeLists.txt代码

在xf_vocie包里的CMakeLists.txt增加以下代码


add_executable(xf_tts src/xf_tts.cpp)
target_link_libraries(xf_tts ${catkin_LIBRARIES} -lmsc  -ldl -lpthread -lm -lrt)
add_dependencies(xf_tts xf_voice_generate_messages_cpp)

至此，我的CMakeLists.txt是这样的

cmake_minimum_required(VERSION 2.8.3)
project(xf_voice)
find_package(catkin REQUIRED COMPONENTS
  roscpp
  rospy
  std_msgs
)
catkin_package()
include_directories(include ${catkin_INCLUDE_DIRS})

add_executable(xf_tts src/xf_tts.cpp)
target_link_libraries(xf_tts ${catkin_LIBRARIES} -lmsc  -ldl -lpthread -lm -lrt)
add_dependencies(xf_tts xf_voice_generate_messages_cpp)

编译

回到catkin_ws目录
然后运行catkin_make

$ catkin_make

调试运行

运行三个终端

第一个终端运行主节点

$ roscore

第二个终端运行语音合成节点

$ cd ~/catkin_ws/
$ source devel/setup.sh
$ rosrun xf_voice xf_tts

第三个终端发布语音信息

$ cd ~/catkin_ws/
$ source devel/setup.sh
$ rostopic list
$ rostopic pub /xfsasywords std_msgs/String 语音合成节点测试

效果图

通过以上测试，你应该已经听到效果了。
运行rqt_graph,可以看到目前的节点图，应该与下图相似，

大吉机器人

关注

1
点赞
踩
13

收藏

觉得还不错? 一键收藏
0
评论
复制链接

分享到 QQ

分享到新浪微博

扫一扫