ros 科大迅飞

ros+科大迅飞语音包+图灵机器人（二）在ros上使用科大迅飞

 

• 在工作空间catkin_ws下创建一个包

• $ cd catkin_ws/src/

• $ catkin_create_pkg voice_system std_msgs rospy roscpp

• 把科大迅飞包里的代码复制到你创建的包里,并重命名为xf_tts.cpp

• $ cd SoftWare/samples/tts_sample/
• $ cp tts_sample.c ~/catkin_ws/src/voice_system/src/
• $ cd catkin_ws/src/voice_system/src/
• $ mv tts_sample.c xf_tts.cpp

• 到SoftWare/include目录下把include拷贝到voice_system包下

• $ cd SoftWare/include
• $ cp * ~/catkin_ws/src/voice_system/include

• 修改xf_tts.cpp文件（注意要把appid改为自己的appid,gtts.h等路径要改为自己该文件的路径）

• xf_tts.cpp的代码如下

• 1. /*

  2. * 语音合成（Text To Speech，TTS）技术能够自动将任意文字实时转换为连续的

  3. * 自然语音，是一种能够在任何时间、任何地点，向任何人提供语音信息服务的

  4. * 高效便捷手段，非常符合信息时代海量数据、动态更新和个性化查询的需求。

  5. */

  6.  

  7. #include <stdlib.h>

  8. #include <stdio.h>

  9. #include <unistd.h>

  10. #include <errno.h>

  11. #include <ros/ros.h>

  12. #include <std_msgs/String.h>

  13. #include "/home/fan/SoftWare/include/qtts.h"

  14. #include "/home/fan/SoftWare/include/msp_cmn.h"

  15. #include "/home/fan/SoftWare/include/msp_errors.h"

  16. const char* fileName="/home/fan/music/voice.wav";

  17. const char* playPath="play /home/fan/music/voice.wav";

  18.  

  19. typedef int SR_DWORD;

  20. typedef short int SR_WORD ;

  21.  

  22. /* wav音频头部格式 */

  23. typedef struct _wave_pcm_hdr

  24. {

  25. char riff[4]; // = "RIFF"

  26. int size_8; // = FileSize - 8

  27. char wave[4]; // = "WAVE"

  28. char fmt[4]; // = "fmt "

  29. int fmt_size; // = 下一个结构体的大小 : 16

  30.  

  31. short int format_tag; // = PCM : 1

  32. short int channels; // = 通道数 : 1

  33. int samples_per_sec; // = 采样率 : 8000 | 6000 | 11025 | 16000

  34. int avg_bytes_per_sec; // = 每秒字节数 : samples_per_sec * bits_per_sample / 8

  35. short int block_align; // = 每采样点字节数 : wBitsPerSample / 8

  36. short int bits_per_sample; // = 量化比特数: 8 | 16

  37.  

  38. char data[4]; // = "data";

  39. int data_size; // = 纯数据长度 : FileSize - 44

  40. } wave_pcm_hdr;

  41.  

  42. /* 默认wav音频头部数据 */

  43. wave_pcm_hdr default_wav_hdr =

  44. {

  45. { 'R', 'I', 'F', 'F' },

  46. 0,

  47. {'W', 'A', 'V', 'E'},

  48. {'f', 'm', 't', ' '},

  49. 16,

  50. 1,

  51. 1,

  52. 16000,

  53. 32000,

  54. 2,

  55. 16,

  56. {'d', 'a', 't', 'a'},

  57. 0

  58. };

  59. /* 文本合成 */

  60. int text_to_speech(const char* src_text, const char* des_path, const char* params)

  61. {

  62. int ret = -1;

  63. FILE* fp = NULL;

  64. const char* sessionID = NULL;

  65. unsigned int audio_len = 0;

  66. wave_pcm_hdr wav_hdr = default_wav_hdr;

  67. int synth_status = MSP_TTS_FLAG_STILL_HAVE_DATA;

  68.  

  69. if (NULL == src_text || NULL == des_path)

  70. {

  71. printf("params is error!\n");

  72. return ret;

  73. }

  74. fp = fopen(des_path, "wb");

  75. if (NULL == fp)

  76. {

  77. printf("open %s error.\n", des_path);

  78. return ret;

  79. }

  80. /* 开始合成 */

  81. sessionID = QTTSSessionBegin(params, &ret);

  82. if (MSP_SUCCESS != ret)

  83. {

  84. printf("QTTSSessionBegin failed, error code: %d.\n", ret);

  85. fclose(fp);

  86. return ret;

  87. }

  88. ret = QTTSTextPut(sessionID, src_text, (unsigned int)strlen(src_text), NULL);

  89. if (MSP_SUCCESS != ret)

  90. {

  91. printf("QTTSTextPut failed, error code: %d.\n",ret);

  92. QTTSSessionEnd(sessionID, "TextPutError");

  93. fclose(fp);

  94. return ret;

  95. }

  96. printf("正在合成 ...\n");

  97. fwrite(&wav_hdr, sizeof(wav_hdr) ,1, fp); //添加wav音频头，使用采样率为16000

  98. while (1)

  99. {

  100. /* 获取合成音频 */

  101. const void* data = QTTSAudioGet(sessionID, &audio_len, &synth_status, &ret);

  102. if (MSP_SUCCESS != ret)

  103. break;

  104. if (NULL != data)

  105. {

  106. fwrite(data, audio_len, 1, fp);

  107. wav_hdr.data_size += audio_len; //计算data_size大小

  108. }

  109. if (MSP_TTS_FLAG_DATA_END == synth_status)

  110. break;

  111. }

  112. printf("\n");

  113. if (MSP_SUCCESS != ret)

  114. {

  115. printf("QTTSAudioGet failed, error code: %d.\n",ret);

  116. QTTSSessionEnd(sessionID, "AudioGetError");

  117. fclose(fp);

  118. return ret;

  119. }

  120. /* 修正wav文件头数据的大小 */

  121. wav_hdr.size_8 += wav_hdr.data_size + (sizeof(wav_hdr) - 8);

  122.  

  123. /* 将修正过的数据写回文件头部,音频文件为wav格式 */

  124. fseek(fp, 4, 0);

  125. fwrite(&wav_hdr.size_8,sizeof(wav_hdr.size_8), 1, fp); //写入size_8的值

  126. fseek(fp, 40, 0); //将文件指针偏移到存储data_size值的位置

  127. fwrite(&wav_hdr.data_size,sizeof(wav_hdr.data_size), 1, fp); //写入data_size的值

  128. fclose(fp);

  129. fp = NULL;

  130. /* 合成完毕 */

  131. ret = QTTSSessionEnd(sessionID, "Normal");

  132. if (MSP_SUCCESS != ret)

  133. {

  134. printf("QTTSSessionEnd failed, error code: %d.\n",ret);

  135. }

  136.  

  137. return ret;

  138. }

  139. int makeTextToWav(const char* text, const char* filename){

  140. int ret = MSP_SUCCESS;

  141. const char* login_params = "appid = 5b090780, work_dir = .";//登录参数,appid与msc库绑定,请勿随意改动

  142. /*

  143. * rdn: 合成音频数字发音方式

  144. * volume: 合成音频的音量

  145. * pitch: 合成音频的音调

  146. * speed: 合成音频对应的语速

  147.  

  148. * voice_name: 合成发音人

  149. * sample_rate: 合成音频采样率

  150. * text_encoding: 合成文本编码格式

  151. *

  152.  

  153. */

  154. const char* session_begin_params = "engine_type = local,voice_name=xiaofeng, text_encoding = UTF8, tts_res_path = fo|res/tts/xiaofeng.jet;fo|res/tts/common.jet, sample_rate = 16000, speed = 50, volume = 50, pitch = 50, rdn = 0";

  155.  

  156. /* 用户登录 */

  157. 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://www.xfyun.cn注册获取

  158. if (MSP_SUCCESS != ret)

  159. {

  160. printf("MSPLogin failed, error code: %d.\n", ret);

  161.  

  162. }

  163. else{

  164. printf("开始合成 ...\n");

  165. ret = text_to_speech(text,filename, session_begin_params);

  166. if (MSP_SUCCESS != ret)

  167. {

  168. printf("text_to_speech failed, error code: %d.\n", ret);

  169. }

  170. printf("合成完毕\n");

  171. }

  172. MSPLogout();

  173. return 0;

  174.  

  175. }

  176. void playWav()

  177. {

  178. system(playPath);

  179.  

  180. }

  181. void topicCallBack(const std_msgs::String::ConstPtr& msg)

  182. {

  183. std::cout<<"get topic text:" << msg->data.c_str();

  184. makeTextToWav(msg->data.c_str(),fileName);

  185. playWav();

  186. }

  187.  

  188. int main(int argc, char* argv[])

  189. { const char* start= "科大迅飞在线语音合成模块启动";

  190. makeTextToWav(start,fileName);

  191. playWav();

  192.  

  193. ros::init(argc,argv, "xf_tts_node");

  194. ros::NodeHandle n;

  195. ros::Subscriber sub = n.subscribe("/voice/xf_tts_topic", 3,topicCallBack);

  196. ros::spin();

  197. return 0;

  198. } 

• 在CMakeList文件的“include_directories”后加入“include”，在文件末尾加入

• 1. add_executable(xf_tts_node src/xf_tts.cpp)

  2. target_link_libraries(xf_tts_node ${catkin_LIBRARIES} -lmsc -lrt -ldl -lpthread)

• CMakeList代码

• 1. cmake_minimum_required(VERSION 2.8.3)

  2. project(voice_system)

  3.  
  4.  

  5. find_package(catkin REQUIRED COMPONENTS

  6. roscpp

  7. rospy

  8. std_msgs

  9. )

  10.  
  11.  

  12. include_directories(

  13. include

  14. ${catkin_INCLUDE_DIRS}

  15.  

  16. )

  17.  
  18.  

  19. add_executable(xf_tts_node src/xf_tts.cpp)

  20. target_link_libraries(xf_tts_node ${catkin_LIBRARIES} -lmsc -lrt -ldl -lpthread)

  21.  
  22.  
  23.  

 

• 在此时到catkin_ws下进行编译
• $ catkin_make
• 编译完后，roscore一下，重新打开窗口到catkin_ws下运行xf_tts_node节点
• $ cd catkin_ws
• $ rosrun voice_system xf_tts_node
• 此时，你能听到，“科大迅飞语音模块启动”的声音“

• 重新打开一个命令窗口，在catkin_ws下发布一个话题
• $ cd catkin_ws
• $ rostopic pub /voice/xf_tts_topic std_msgs/String 你好
• 此时，你会听到”你好的声音“

• 证明ros上运行科大迅飞语音模块成功