Respeaker multi channel fetch15

3rdparty/voice_text/CMakeLists.txt

@@ -28,32 +28,51 @@ 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()
+
+# 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})
+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(
+  ${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)
 
 if(NOT VT_LIB_PATH)
   message(WARNING "Building dummy library")
@@ -64,10 +83,11 @@ if(NOT VT_LIB_PATH)
   set(VT_LIB_PATH ${PROJECT_BINARY_DIR}/libvt_jpn.so)
 endif()
 
-  target_link_libraries(voice_text
+target_link_libraries(voice_text
     ${catkin_LIBRARIES}
     ${VT_LIB_PATH} -lm -lpthread
-  )
+    ${VT_API_LIB_PATH} -lm -lpthread -ldl
+)
 if(NOT EXISTS ${VT_LIB_PATH})
   add_dependencies(voice_text vt_dummy)
 endif()

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/launch/voice_text.launch

@@ -9,7 +9,8 @@
 
   <arg name="sound_play_respawn" default="true" />
 
-  <node name="voice_text" pkg="voice_text" type="voice_text" output="screen"
+  <node name="voice_text" pkg="voice_text" type="voice_text"
+        output="screen" respawn="true" respawn_delay="10"
         machine="$(arg voice_text_machine)">
     <remap from="text_to_speech" to="voice_text/text_to_speech" />
     <rosparam>

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

@@ -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
@@ -45,7 +45,13 @@ public:
 
   ~VoiceText() {
     if (initialized_) {
+      #ifdef _VTAPI_H_
+      VTAPI_ReleaseHandle(this->hVTAPI);
+      VTAPI_UnloadEngine(this->hEngine);
+      VTAPI_Exit();
+      #else
       VT_UNLOADTTS_JPN(-1);
+      #endif
     }
   }
 
@@ -64,17 +70,37 @@ public:
       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_ROOT@/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) {
         ROS_FATAL("You must install voice_text library before building this library");
       }
       return false;
     }
+    VT_GetTTSInfo_JPN(VT_VERIFY_CODE, NULL, &ret, sizeof(int));
+    if (ret != 0) {
+      ROS_FATAL_STREAM("Verification failed (VT_VERIFY_CODE " << ret << ")");
+      return false;
+    }
 
+    // Print voice text info
     char szTmp[PATH_MAX];
     char szTmp2[PATH_MAX];
     VT_GetTTSInfo_JPN(VT_BUILD_DATE, NULL, szTmp, PATH_MAX);
@@ -95,6 +121,10 @@ public:
     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);
@@ -124,6 +154,27 @@ public:
     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,
@@ -138,12 +189,12 @@ public:
 
     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;
@@ -156,6 +207,10 @@ public:
   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) {

3rdparty/voicevox/.gitignore

@@ -0,0 +1,6 @@
+build
+dict
+lib
+node_scripts/voicevox_engine
+requirements.txt
+!.gitignore

3rdparty/voicevox/CMakeLists.txt

@@ -0,0 +1,73 @@
+cmake_minimum_required(VERSION 2.8.3)
+project(voicevox)
+
+find_package(catkin REQUIRED
+  COMPONENTS
+  catkin_virtualenv
+)
+
+catkin_python_setup()
+
+set(INSTALL_DIR ${PROJECT_SOURCE_DIR})
+
+catkin_package()
+
+catkin_generate_virtualenv(
+  INPUT_REQUIREMENTS requirements.in
+  PYTHON_INTERPRETER python3
+  USE_SYSTEM_PACKAGES FALSE
+)
+
+add_custom_command(
+  OUTPUT voicevox_model_installed
+  COMMAND make -f ${PROJECT_SOURCE_DIR}/Makefile.model
+  MD5SUM_DIR=${PROJECT_SOURCE_DIR}/md5sum
+  INSTALL_DIR=${INSTALL_DIR}
+)
+
+
+add_custom_command(
+  OUTPUT voicevox_core_installed
+  COMMAND make -f ${PROJECT_SOURCE_DIR}/Makefile.core
+  MD5SUM_DIR=${PROJECT_SOURCE_DIR}/md5sum
+  INSTALL_DIR=${INSTALL_DIR}
+)
+
+add_custom_command(
+  OUTPUT voicevox_engine_installed
+  COMMAND make -f ${PROJECT_SOURCE_DIR}/Makefile.engine
+  MD5SUM_DIR=${PROJECT_SOURCE_DIR}/md5sum
+  INSTALL_DIR=${INSTALL_DIR}
+)
+
+add_custom_command(
+  OUTPUT open_jtalk_dic_installed
+  COMMAND make -f ${PROJECT_SOURCE_DIR}/Makefile.open_jtalk_dic
+  MD5SUM_DIR=${PROJECT_SOURCE_DIR}/md5sum
+  INSTALL_DIR=${INSTALL_DIR}
+)
+
+add_custom_target(all_installed ALL DEPENDS
+  voicevox_model_installed
+  voicevox_core_installed
+  voicevox_engine_installed
+  open_jtalk_dic_installed)
+
+file(GLOB NODE_SCRIPTS_FILES node_scripts/*.py)
+catkin_install_python(
+  PROGRAMS ${NODE_SCRIPTS_FILES}
+  DESTINATION ${CATKIN_PACKAGE_BIN_DESTINATION}/node_scripts/
+)
+install(DIRECTORY node_scripts/voicevox_engine
+  DESTINATION ${CATKIN_PACKAGE_SHARE_DESTINATION}/catkin_virtualenv_scripts/
+  USE_SOURCE_PERMISSIONS)
+install(DIRECTORY launch dict
+  DESTINATION ${CATKIN_PACKAGE_SHARE_DESTINATION}
+  USE_SOURCE_PERMISSIONS)
+install(PROGRAMS bin/text2wave
+  DESTINATION ${CATKIN_PACKAGE_SHARE_DESTINATION}/bin)
+
+install(DIRECTORY
+  ${INSTALL_DIR}/lib
+  DESTINATION ${CATKIN_PACKAGE_SHARE_DESTINATION}
+  USE_SOURCE_PERMISSIONS)

3rdparty/voicevox/Makefile

@@ -0,0 +1,11 @@
+all:
+	make -f Makefile.core
+	make -f Makefile.model
+	make -f Makefile.engine
+	make -f Makefile.open_jtalk_dic
+clean:
+	make -f Makefile.core clean
+	make -f Makefile.model clean
+	make -f Makefile.engine clean
+	make -f Makefile.open_jtalk_dic clean
+	rm -rf build

3rdparty/voicevox/Makefile.core

@@ -0,0 +1,28 @@
+# -*- makefile -*-
+
+all: installed.viocevox_core
+
+VERSION = 0.11.4
+FILENAME = core.zip
+TARBALL = build/$(FILENAME)
+TARBALL_URL = "https://github.com/VOICEVOX/voicevox_core/releases/download/$(VERSION)/core.zip"
+SOURCE_DIR = build/core
+UNPACK_CMD = unzip
+MD5SUM_DIR = $(CURDIR)/md5sum
+MD5SUM_FILE = $(MD5SUM_DIR)/$(FILENAME).md5sum
+SCRIPT_DIR = $( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )
+include $(shell rospack find mk)/download_unpack_build.mk
+INSTALL_DIR =	'./'
+
+
+installed.viocevox_core: $(SOURCE_DIR)/unpacked
+	mkdir -p $(INSTALL_DIR)/lib
+	cp build/core/lib*.so $(INSTALL_DIR)/lib/
+	cp build/core/*.bin $(INSTALL_DIR)/lib/
+	cp build/core/metas.json $(INSTALL_DIR)/lib/metas.json
+
+clean:
+	rm -rf $(TARBALL)
+	rm -rf $(SOURCE_DIR)
+	rm -rf $(INSTALL_DIR)/lib
+	rm -rf build

3rdparty/voicevox/Makefile.engine

@@ -0,0 +1,24 @@
+# -*- makefile -*-
+
+all: installed.voicevox_engine
+
+VERSION = 0.11.4
+FILENAME = $(VERSION).tar.gz
+TARBALL = build/$(FILENAME)
+TARBALL_URL = "https://github.com/VOICEVOX/voicevox_engine/archive/refs/tags/$(FILENAME)"
+SOURCE_DIR = build/voicevox_engine-$(VERSION)
+UNPACK_CMD = tar xvzf
+MD5SUM_DIR = $(CURDIR)/md5sum
+MD5SUM_FILE = $(MD5SUM_DIR)/voicevox_engine.tar.gz.md5sum
+include $(shell rospack find mk)/download_unpack_build.mk
+INSTALL_DIR =	'./'
+
+
+installed.voicevox_engine: $(SOURCE_DIR)/unpacked
+	cp -r build/voicevox_engine-$(VERSION) $(INSTALL_DIR)/node_scripts/voicevox_engine
+
+clean:
+	rm -rf $(TARBALL)
+	rm -rf $(SOURCE_DIR)
+	rm -rf $(INSTALL_DIR)/node_scripts/voicevox_engine
+	rm -rf build