（二）說話篇

基本環境配置

建立工作空間

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

• 首次編譯

$ cd ~/catkin_ws/
$ catkin_make

• 建立語音包

$ cd src/
$ catkin_create_pkg xf_voice std_msgs rospy roscpp

安裝語音庫

• 建立/Robot資料夾將所有者改為當前使用者

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

• 然後將你下載的語音sdk中的libmsc.so放到/Robot/voice/lib/資料夾下
• 將bin目錄和lib/inc目錄都放到/Robot/voice/資料夾下，
• 安裝mpalyer播放器
• 建立/Robot/voice/wav和/Robot/cmd資料夾，前者用來儲存臨時音訊檔案，後者用來儲存管道通訊檔案。

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

• 至此你的/Robot資料夾應該有cmd和voice兩個資料夾，voice下應該有bin inc lib wav四個資料夾
• 然後我們把語音庫放到系統庫資料夾/usr/lib/

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

編寫語音合成節點

原始碼

• 回到工作空間目錄，~/catkin_ws
• 在~/catkin_ws/src/xf_voice/src目錄下新建檔案xf_tts.cpp 並將以下內容複製進去
• 注意：將裡面的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音訊頭部格式 */
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;               // = 下一個結構體的大小 : 16

    short int       format_tag;             // = PCM : 1
    short int       channels;               // = 通道數 : 1
    int             samples_per_sec;        // = 取樣率 : 8000 | 6000 | 11025 | 16000
    int             avg_bytes_per_sec;      // = 每秒位元組數 : samples_per_sec * bits_per_sample / 8
    short int       block_align;            // = 每取樣點位元組數 : wBitsPerSample / 8
    short int       bits_per_sample;        // = 量化位元數: 8 | 16

    char            data[4];                // = "data";
    int             data_size;              // = 純資料長度 : FileSize - 44 
} wave_pcm_hdr;
/* 預設wav音訊頭部資料 */
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  
};
/* 文字合成 */
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;
    }
    /* 開始合成 */
    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); //新增wav音訊頭，使用取樣率為16000
    while (1) 
    {
        /* 獲取合成音訊 */
        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; //計算data_size大小
        }
        if (MSP_TTS_FLAG_DATA_END == synth_status)
            break;
    }//合成狀態synth_status取值請參閱《訊飛語音雲API文件》
    printf("\n");
    if (MSP_SUCCESS != ret)
    {
        printf("QTTSAudioGet failed, error code: %d.\n",ret);
        QTTSSessionEnd(sessionID, "AudioGetError");
        fclose(fp);
        return ret;
    }
    /* 修正wav檔案頭資料的大小 */
    wav_hdr.size_8 += wav_hdr.data_size + (sizeof(wav_hdr) - 8);
    /* 將修正過的資料寫回檔案頭部,音訊檔案為wav格式 */
    fseek(fp, 4, 0);
    fwrite(&wav_hdr.size_8,sizeof(wav_hdr.size_8), 1, fp); //寫入size_8的值
    fseek(fp, 40, 0); //將檔案指標偏移到儲存data_size值的位置
    fwrite(&wav_hdr.data_size,sizeof(wav_hdr.data_size), 1, fp); //寫入data_size的值
    fclose(fp);
    fp = NULL;
    /* 合成完畢 */
    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 = .";//登入引數,appid與msc庫繫結,請勿隨意改動
    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";
    /* 使用者登入 */
    ret = MSPLogin(NULL, NULL, login_params); //第一個引數是使用者名稱，第二個引數是密碼，第三個引數是登入引數，使用者名稱和密碼可在http://open.voicecloud.cn註冊獲取
    if (MSP_SUCCESS != ret)
    {
        printf("MSPLogin failed, error code: %d.\n", ret);
        goto exit ;//登入失敗，退出登入
    }
    /* 文字合成 */
    printf("開始合成 ...\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("合成完畢\n");

exit:
    MSPLogout(); //退出登入
    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"; //合成的語音檔名稱
    const char* text                 = "語音合成模組啟動成功！"; //合成文字
    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程式碼

• 在xf_vocie包裡的CMakeLists.txt增加以下程式碼

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是這樣的

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)

編譯

• 回到catkin_ws目錄
• 然後執行catkin_make

$ catkin_make

除錯執行

執行三個終端

• 第一個終端 執行主節點

$ roscore

• 第二個終端 執行語音合成節點

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

• 第三個終端 釋出語音資訊

$ cd ~/catkin_ws/
$ source devel/setup.sh
$ rostopic list
$ rostopic pub /xfsasywords std_msgs/String 語音合成節點測試

效果圖

• 通過以上測試，你應該已經聽到效果了。
  
• 執行rqt_graph,可以看到目前的節點圖，應該與下圖相似，