Merge pull request #395 from mqcmd196/PR/merge_vtapi_vtsdk

[voice_text] Support new ReadSpeaker API and load libraries dynamically

Author: k-okada

3rdparty/voice_text/.gitignore

@@ -1,5 +1,25 @@
 cmake_minimum_required(VERSION 2.8.3)
 project(voice_text)
+
+# voice_text requires C++ 11
+if(CMAKE_VERSION VERSION_LESS "3.1")
+  if(NOT CMAKE_CXX_FLAGS MATCHES "-std=")
+    include(CheckCXXCompilerFlag)
+    check_cxx_compiler_flag("-std=c++11" HAS_CXX11_FLAG)
+    if(HAS_CXX11_FLAG)
+      set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
+    else()
+      check_cxx_compiler_flag("-std=gnu++11" HAS_GNUXX11_FLAG)
+      if(HAS_GNUXX11_FLAG)
+        set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=gnu++11")
+      endif()
+    endif()
+  endif()
+endif()
+if("$ENV{ROS_DISTRO}" STREQUAL "kinetic")
+  add_compile_options(-std=c++11)
+endif()
+
 find_package(catkin REQUIRED COMPONENTS
   dynamic_reconfigure
   roscpp
@@ -20,64 +40,28 @@ generate_messages()
 
 catkin_package(CATKIN_DEPENDS message_runtime)
 
-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")
-  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 ${VT_ROOT}/bin/LINUX32_GLIBC3/RAMIO/libvt_jpn.so)  # e.g., /usr/vt/risa/H16/bin/LINUX32_GLIBC3/RAMIO/libvt_jpn.so
-  elseif("${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
-  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()
-  endif()
-  if(VT_LIB_PATH)
-    message(WARNING "VoiceText 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}")
-  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)
+include_directories(
+  include
+  ${Boost_INCLUDE_DIRS}
+  ${catkin_INCLUDE_DIRS}
+)
 
-if(NOT VT_LIB_PATH)
-  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)
-endif()
+add_executable(voice_text src/voice_text.cpp src/vt_handler.cpp)
+add_dependencies(voice_text ${PROJECT_NAME}_generate_messages_cpp ${PROJECT_NAME}_gencfg)
+set_target_properties(voice_text PROPERTIES COMPILE_FLAGS -D_REENTRANT)
 
-  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()
+target_link_libraries(voice_text
+    ${catkin_LIBRARIES} -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}
   LIBRARY DESTINATION ${CATKIN_PACKAGE_LIB_DESTINATION}
   RUNTIME DESTINATION ${CATKIN_PACKAGE_BIN_DESTINATION}
 )
-
+catkin_install_python(PROGRAMS scripts/text2wave.py
+  DESTINATION ${CATKIN_PACKAGE_BIN_DESTINATION}
+)
 install(PROGRAMS bin/text2wave
   DESTINATION ${CATKIN_PACKAGE_SHARE_DESTINATION}/bin
 )

3rdparty/voice_text/CMakeLists.txt

@@ -5,9 +5,20 @@ 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 ReadSpeaker 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. 
+  ```
+### 2. Put license file
+### 3. Build this package
 
 ```bash
 cd /path/to/catkin_workspace
@@ -55,7 +66,7 @@ Robot says "Hello!"
 
 ### Parameters
 
-* `~db_path` (String, default: path found at build time (e.g., `"/var/vt/sayaka/M16"`))
+* `~db_path` (String, default: vt path (e.g., `"/var/vt/sayaka/M16"`))
 
   Path to VoiceText database directory.
 

3rdparty/voice_text/README.md

@@ -1,43 +1,3 @@
-#!/usr/bin/env python
+#!/usr/bin/env sh
 
-import argparse
-import sys
-import os
-import rospy
-from voice_text.srv import *
-
-
-def text_to_speech_client(text_path, wave_path):
-    rospy.wait_for_service('/voice_text/text_to_speech')
-    try:
-        text_to_speech = rospy.ServiceProxy('/voice_text/text_to_speech', TextToSpeech)
-        # print("Requesting %s -> %s"%(text_path, wave_path))
-        res = text_to_speech(text_path, wave_path)
-        return res.ok
-    except rospy.ServiceException as e:
-        print("Service call failed: %s"%e)
-
-
-if __name__ == '__main__':
-    parser = argparse.ArgumentParser(description='')
-    parser.add_argument('-eval', '--evaluate')
-    parser.add_argument('-o', '--output')
-    parser.add_argument('text')
-    args = parser.parse_args()
-
-    input_file = args.text
-    jptext_file = "/tmp/_voice_text_%s.txt"%os.getpid()
-    output_file = "/tmp/_voice_text_%s.wav"%os.getpid()
-    if args.output is not None:
-        output_file = args.output
-
-    if not os.path.exists(input_file):
-        rospy.logerr("%s not found"%input_file)
-        sys.exit(1)
-
-    # convert character code to shift JIS
-    os.system("nkf -s %s > %s"%(input_file, jptext_file))
-
-    text_to_speech_client(jptext_file, output_file)
-
-    os.remove(jptext_file)
+exec rosrun voice_text text2wave.py $@