Support new VoiceText API

3rdparty/voice_text/CMakeLists.txt

@@ -20,57 +20,81 @@ generate_messages()
 
 catkin_package(CATKIN_DEPENDS message_runtime)
 
+# find voice text headers and libraries
 file(GLOB VT_ROOT /usr/vt/*/*)
 if(NOT VT_ROOT)
-  message(WARNING "VoiceText directory should be /usr/vt/*/* (e.g., /usr/vt/sayaka/M16) but is not found")
+  message(WARNING "VoiceText directory should be /usr/vt/*/* (e.g., /usr/vt/sayaka/M16) but it is not found")
   set(VT_ROOT /usr/vt/sayaka/M16)  # default value for following configure_file
 else()
   if("${CMAKE_SIZEOF_VOID_P}" STREQUAL "4")
     set(VT_LIB_PATH_OLD ${VT_ROOT}/bin/x86_32/RAMIO/libvt_jpn.so)  # e.g., /usr/vt/sayaka/M16/bin/x86_32/RAMIO/libvt_jpn.so
     set(VT_LIB_PATH_NEW ${VT_ROOT}/bin/LINUX32_GLIBC3/RAMIO/libvt_jpn.so)  # e.g., /usr/vt/risa/H16/bin/LINUX32_GLIBC3/RAMIO/libvt_jpn.so
+    set(VT_LIB_PATH_FILEIO ${VT_ROOT}/bin/FILEIO/LINUX32_GLIBC3/libvt_jpn.so)  # e.g., /usr/vt/hikari/D16/bin/FILEIO/LINUX32_GLIBC3/libvt_jpn.so
   elseif("${CMAKE_SIZEOF_VOID_P}" STREQUAL "8")
     set(VT_LIB_PATH_OLD ${VT_ROOT}/bin/x86_64/RAMIO/libvt_jpn.so)  # e.g., /usr/vt/sayaka/M16/bin/x86_64/RAMIO/libvt_jpn.so
     set(VT_LIB_PATH_NEW ${VT_ROOT}/bin/LINUX64_GLIBC3/RAMIO/libvt_jpn.so)  # e.g., /usr/vt/risa/H16/bin/LINUX64_GLIBC3/RAMIO/libvt_jpn.so
+    set(VT_LIB_PATH_FILEIO ${VT_ROOT}/bin/FILEIO/LINUX64_GLIBC3/libvt_jpn.so)  # e.g., /usr/vt/hikari/D16/bin/FILEIO/LINUX64_GLIBC3/libvt_jpn.so
   endif()
   if(EXISTS ${VT_LIB_PATH_OLD})
     set(VT_LIB_PATH ${VT_LIB_PATH_OLD})
-  else()
-    if(EXISTS ${VT_LIB_PATH_NEW})
-      set(VT_LIB_PATH ${VT_LIB_PATH_NEW})
-    endif()
+  elseif(EXISTS ${VT_LIB_PATH_NEW})
+    set(VT_LIB_PATH ${VT_LIB_PATH_NEW})
+  elseif(EXISTS ${VT_LIB_PATH_FILEIO})
+    set(VT_LIB_PATH ${VT_LIB_PATH_FILEIO})
   endif()
   if(VT_LIB_PATH)
-    message(WARNING "VoiceText library is found at ${VT_LIB_PATH}")
+    message(WARNING "VoiceText Engine library is found at ${VT_LIB_PATH}")
   else()
-    message(WARNING "VoiceText library is not found at ${VT_LIB_PATH_OLD} or ${VT_LIB_PATH_NEW}")
+    message(WARNING "VoiceText Engine library is not found at ${VT_LIB_PATH_OLD} or ${VT_LIB_PATH_NEW} or ${VT_LIB_PATH_FILE}")
   endif()
 endif()
-configure_file(src/voice_text.cpp.in ${PROJECT_SOURCE_DIR}/src/voice_text.cpp)
 
-  include_directories(
-    ${Boost_INCLUDE_DIRS}
-    ${catkin_INCLUDE_DIRS}
-  )
-  add_executable(voice_text src/voice_text.cpp)
-  add_dependencies(voice_text ${PROJECT_NAME}_generate_messages_cpp ${PROJECT_NAME}_gencfg)
-  set_target_properties(voice_text PROPERTIES COMPILE_FLAGS -D_REENTRANT)
+# use traditional sdk or new api
+set(VT_SDK_H ${VT_ROOT}/inc/vt_jpn.h)
+set(VT_API_H ${VT_ROOT}/inc/vtapi.h)
+if(EXISTS ${VT_SDK_H})
+  # use traditional sdk
+  message(WARNING "Using SDK because ${VT_SDK_H} was found.")
+  set(VT_INCLUDE ${VT_SDK_H})
+elseif(EXISTS ${VT_API_H})
+  # use new api
+  message(WARNING "Using API because ${VT_SDK_H} was not found but ${VT_API_H} was found.")
+  string(REPLACE "/" ";" _VT_ROOT_SPLIT ${VT_ROOT})
+  list(GET _VT_ROOT_SPLIT 3 VT_SPEAKER)
+  list(GET _VT_ROOT_SPLIT 4 VT_TYPE)
+  set(VT_API_LIB_PATH ${VT_ROOT}/bin/libvtapi.so)
+  set(VT_INCLUDE ${VT_API_H})
+  set(VT_BIN ${VT_ROOT}/bin)
+endif()
 
-if(NOT VT_LIB_PATH)
+include_directories(
+  ${Boost_INCLUDE_DIRS}
+  ${catkin_INCLUDE_DIRS}
+)
+add_executable(voice_text src/voice_text.cpp)
+add_dependencies(voice_text ${PROJECT_NAME}_generate_messages_cpp ${PROJECT_NAME}_gencfg)
+set_target_properties(voice_text PROPERTIES COMPILE_FLAGS -D_REENTRANT)
+
+# build with voice_text engint or dummy
+if(VT_LIB_PATH)
+  message(WARNING "Building with Voice Text library")
+  set(BUILD_FLAGS "-DVT_ROOT=${VT_ROOT} -DVT_INCLUDE=${VT_INCLUDE} -DVT_SPEAKER=${VT_SPEAKER} -DVT_TYPE=${VT_TYPE}")
+  set_target_properties(voice_text PROPERTIES COMPILE_OPTIONS ${BUILD_FLAGS})
+else()
   message(WARNING "Building dummy library")
   add_library(vt_dummy src/dummy/vt_dummy.cpp)
   set_target_properties(vt_dummy PROPERTIES LIBRARY_OUTPUT_DIRECTORY ${PROJECT_BINARY_DIR})
   set_target_properties(vt_dummy PROPERTIES LIBRARY_OUTPUT_NAME vt_jpn)
   set_target_properties(voice_text PROPERTIES COMPILE_FLAGS -DUSE_DUMMY_INCLUDE)
   set(VT_LIB_PATH ${PROJECT_BINARY_DIR}/libvt_jpn.so)
+  add_dependencies(voice_text vt_dummy)
 endif()
 
-  target_link_libraries(voice_text
+target_link_libraries(voice_text
     ${catkin_LIBRARIES}
     ${VT_LIB_PATH} -lm -lpthread
-  )
-if(NOT EXISTS ${VT_LIB_PATH})
-  add_dependencies(voice_text vt_dummy)
-endif()
+    ${VT_API_LIB_PATH} -lm -lpthread -ldl
+)
 
 install(TARGETS voice_text # do not install vt_dummy target, that should be installed from voice_text library
   ARCHIVE DESTINATION ${CATKIN_PACKAGE_LIB_DESTINATION}

3rdparty/voice_text/README.md

@@ -5,9 +5,23 @@ ROS Interface for HOYA VoiceText Speech Synthesis Engine
 
 ## Installation
 
-1. Install VoiceText SDK
-2. Put license file
-3. Build this package
+### 1. Install VoiceText SDK
+#### If you have voicetext sdk install binary, please follow the official guide and install both engine and SDK
+#### If you don't have the sdk install binary but have voice text API binary, please follow the guide below.
+1. Install VoiceText Engine by official guide
+2. Copy VoiceText API binaries to VoiceText binary directory
+  VoiceText API package includes binary libraries and header file. You have to copy those of them to specific directory by executing following commands.
+  ```bash
+  cd /path_to_api_package_directory # e.g. cd ~/Downloads/RS_VTAPI_SDK_Linux_4.3.0.2/20201113_VTAPI4.3.0.2_LINUX
+  cd bin/x64 # You have to cd x86 if your system is x86 architecture
+  # Assuming VoiceText engine's talker is hikari, type is D16. If it is different, please set appropriate directory.
+  sudo cp -a * /usr/vt/hikari/D16/bin # Don't forget to add -a not to break symbolic link. 
+  cd ../../include/
+  sudo mkdir /usr/vt/hikari/D16/inc # not include, but inc
+  sudo cp vtapi.h /usr/vt/hikari/D16/inc
+  ```
+### 2. Put license file
+### 3. Build this package
 
 ```bash
 cd /path/to/catkin_workspace

3rdparty/voice_text/src/voice_text.cpp.in → 3rdparty/voice_text/src/voice_text.cpp

@@ -21,7 +21,7 @@
 #ifdef USE_DUMMY_INCLUDE
 #include "dummy/vt_dummy.h"
 #else
-#include "@VT_ROOT@/inc/vt_jpn.h"
+#include VT_INCLUDE
 #endif
 
 #define PATH_MAX 1024
@@ -34,7 +34,11 @@ class VoiceText {
   typedef voice_text::VoiceTextConfig Config;
 
   VoiceText() : nh_(), pnh_("~"), db_path_(""), license_path_(""), dyn_srv_(pnh_) {
-    pnh_.param<std::string>("db_path", db_path_, "@VT_ROOT@");
+#ifdef USE_DUMMY_INCLUDE
+    pnh_.param<std::string>("db_path", db_path_, "dummy");
+#else
+    pnh_.param<std::string>("db_path", db_path_, VT_ROOT);
+#endif
     pnh_.setParam("db_path", db_path_);  // for backward compatibility (db_path is usually set previously)
     pnh_.param<std::string>("license_path", license_path_, "");
 
@@ -45,7 +49,13 @@ class VoiceText {
 
   ~VoiceText() {
     if (initialized_) {
+      #ifdef _VTAPI_H_
+      VTAPI_ReleaseHandle(this->hVTAPI);
+      VTAPI_UnloadEngine(this->hEngine);
+      VTAPI_Exit();
+      #else
       VT_UNLOADTTS_JPN(-1);
+      #endif
     }
   }
 
@@ -64,9 +74,23 @@ class VoiceText {
       license_path_char = (char*)calloc(std::strlen(license_path_.c_str())+1, sizeof(char));
       std::strcpy(license_path_char, license_path_.c_str());
     }
+    #ifdef _VTAPI_H_
+    VTAPI_Init(VT_BIN); // set voice_text binary library path because it loads them dynamically
+    this->hVTAPI = VTAPI_CreateHandle();
+    if (this->hVTAPI == 0) {
+      ROS_ERROR("VoiceText API ERROR when creating API handler. : %s\n", VTAPI_GetLastErrorInfo(0)->szMsg);
+      return false;
+    }
+    VTAPI_SetLicenseFolder(license_path_char);
+    // Load engine
+    this->hEngine = VTAPI_GetEngine(VT_SPEAKER, VT_TYPE);
+    ret = VTAPI_SetEngineHandle(this->hVTAPI, this->hEngine);
+    if(ret < VTAPI_SUCCESS){
+      ROS_ERROR("VoiceText API ERROR when creating engine handler. CODE: %d, MESSAGE: %s \n", ret, VTAPI_GetLastErrorInfo(this->hVTAPI)->szMsg);
+      return false;
+    }
+    #else
     ret = VT_LOADTTS_JPN((int)NULL, -1, db_path_char, license_path_char);
-    free(db_path_char);
-    if (!license_path_.empty()) free(license_path_char);
     if (ret != VT_LOADTTS_SUCCESS) {
       ROS_FATAL("Failed to load TTS engine (code %d)", ret);
       if (ret == -1) {
@@ -101,6 +125,10 @@ class VoiceText {
     VT_GetTTSInfo_JPN(VT_CODEPAGE, NULL, &ret, sizeof(int));
     sprintf(szTmp2, "VT_GetTTSInfo_JPN(VT_CODEPAGE = %d\n", ret);
     ROS_INFO_STREAM(szTmp2);
+    #endif
+
+    free(db_path_char);
+    if (!license_path_.empty()) free(license_path_char);
 
     // advertise service
     srv_ = nh_.advertiseService("text_to_speech", &VoiceText::text_to_speech, this);
@@ -130,6 +158,27 @@ class VoiceText {
     char* wave_char = (char*)calloc(std::strlen(req.wave_path.c_str())+1, sizeof(char));
     std::strcpy(wave_char, req.wave_path.c_str());
 
+    #ifdef _VTAPI_H_
+    int ret;
+    // set attributes
+    VTAPI_SetAttr(this->hVTAPI, ATTR_PITCH, config_.pitch);
+    VTAPI_SetAttr(this->hVTAPI, ATTR_SPEED, config_.speed);
+    VTAPI_SetAttr(this->hVTAPI, ATTR_VOLUME, config_.volume);
+    VTAPI_SetAttr(this->hVTAPI, ATTR_PAUSE, config_.pause);
+    ret = VTAPI_SetOutputFile(this->hVTAPI, wave_char, FORMAT_16PCM_WAV);
+    if(ret != VTAPI_SUCCESS){
+      ROS_ERROR("VoiceText API ERROR when executing VTAPI_SetOutputFile. STATUS: %s\n", VTAPI_GetLastErrorInfo(this->hVTAPI)->szMsg);
+      res.ok = false;
+    }
+    // set input text file path
+    ret = VTAPI_TextToFile(this->hVTAPI, text_char, -1, TEXT_FORMAT_DEFAULT);
+    if(ret != VTAPI_SUCCESS){
+      ROS_ERROR("VoiceText API ERROR when executing VTAPI_TextToFile. STATUS: %s\n", VTAPI_GetLastErrorInfo(this->hVTAPI)->szMsg);
+      res.ok = false;
+    }
+    free(text_char);
+    free(wave_char);
+    #else
     int ret = VT_TextToFile_JPN(VT_FILE_API_FMT_S16PCM_WAVE,
                                 text_char,
                                 wave_char,
@@ -144,12 +193,12 @@ class VoiceText {
 
     free(text_char);
     free(wave_char);
-
     if (ret != VT_FILE_API_SUCCESS) {
       ROS_ERROR("Failed to execute tts: (code: %d)", ret);
       res.ok = false;
       return true;
     }
+    #endif
 
     res.ok = true;
     return true;
@@ -162,6 +211,10 @@ class VoiceText {
   ros::ServiceServer srv_;
   bool initialized_;
   std::string db_path_, license_path_;
+  #ifdef _VTAPI_H_
+  VTAPI_HANDLE hVTAPI;
+  VTAPI_ENGINE_HANDLE hEngine;
+  #endif
 };
 
 int main(int argc, char** argv) {