feat(stt): add NVIDIA Canary STT engine support

CMakeLists.txt

@@ -374,6 +374,8 @@ set(dsnote_lib_sources
     ${sources_dir}/app_server.cpp
     ${sources_dir}/fasterwhisper_engine.hpp
     ${sources_dir}/fasterwhisper_engine.cpp
+    ${sources_dir}/canary_engine.hpp
+    ${sources_dir}/canary_engine.cpp
     ${sources_dir}/mimic3_engine.hpp
     ${sources_dir}/mimic3_engine.cpp
     ${sources_dir}/april_engine.hpp
@@ -533,17 +535,17 @@ pkg_search_module(pulse REQUIRED libpulse)
 list(APPEND deps_libs ${pulse_LIBRARIES})
 list(APPEND includes ${pulse_INCLUDE_DIRS})
 
-find_package(Qt5 COMPONENTS Core Network Multimedia Qml Xml Sql Gui Quick DBus LinguistTools REQUIRED)
-list(APPEND deps_libs Qt5::Core Qt5::Network Qt5::Multimedia Qt5::Gui Qt5::Quick Qt5::DBus Qt5::Xml)
+find_package(Qt6 COMPONENTS Core Network Multimedia Qml Xml Sql Gui Quick DBus LinguistTools REQUIRED)
+list(APPEND deps_libs Qt6::Core Qt6::Network Qt6::Multimedia Qt6::Gui Qt6::Quick Qt6::DBus Qt6::Xml)
 
 if(WITH_DESKTOP)
-    find_package(Qt5 COMPONENTS QuickControls2 Widgets REQUIRED)
-    list(APPEND deps_libs Qt5::QuickControls2 Qt5::Widgets)
+    find_package(Qt6 COMPONENTS QuickControls2 Widgets REQUIRED)
+    list(APPEND deps_libs Qt6::QuickControls2 Qt6::Widgets)
 
     pkg_search_module(wayland REQUIRED wayland-client)
     list(APPEND deps_libs ${wayland_LIBRARIES})
     list(APPEND includes ${wayland_INCLUDE_DIRS})
-    include_directories(${Qt5Gui_PRIVATE_INCLUDE_DIRS})
+    include_directories(${Qt6Gui_PRIVATE_INCLUDE_DIRS})
 
     if(BUILD_XKBCOMMON)
         include(${cmake_path}/xkbcommon.cmake)
@@ -565,8 +567,10 @@ if(WITH_DESKTOP)
         find_package(X11 REQUIRED)
         list(APPEND deps_libs X11)
 
-        find_package(Qt5 COMPONENTS X11Extras REQUIRED)
-        list(APPEND deps_libs Qt5::X11Extras)
+        # Qt6: X11Extras removed, functionality now in QtGui via QNativeInterface
+        pkg_search_module(xcb REQUIRED xcb)
+        list(APPEND deps_libs ${xcb_LIBRARIES})
+        list(APPEND includes ${xcb_INCLUDE_DIRS})
 
         if(BUILD_QHOTKEY)
             include(${cmake_path}/qhotkey.cmake)

cmake/dbus_api.cmake

@@ -2,12 +2,12 @@ set(dbus_dsnote_interface_file "${PROJECT_BINARY_DIR}/${info_dbus_app_interface}
 
 configure_file(${dbus_dir}/dsnote.xml.in ${dbus_dsnote_interface_file})
 
-find_package(Qt5 COMPONENTS DBus REQUIRED)
+find_package(Qt6 COMPONENTS DBus REQUIRED)
 
 unset(qdbusxml2cpp_bin CACHE)
-find_program(qdbusxml2cpp_bin qdbusxml2cpp)
+find_program(qdbusxml2cpp_bin qdbusxml2cpp HINTS ${Qt6_DIR}/../../../libexec ${Qt6_DIR}/../../../bin)
 if(${qdbusxml2cpp_bin} MATCHES "-NOTFOUND$")
-   find_program(qdbusxml2cpp_bin qdbusxml2cpp-qt5)
+   find_program(qdbusxml2cpp_bin qdbusxml2cpp-qt6)
    if(${qdbusxml2cpp_bin} MATCHES "-NOTFOUND$")
       message(FATAL_ERROR "qdbusxml2cpp not found but it is required")
    endif()

cmake/openblas_pkgconfig.cmake

@@ -1,6 +1,6 @@
 pkg_search_module(openblas openblas)
 
-if(NOT DEFINED ${openblas_FOUND})
+if(NOT openblas_FOUND)
     # check without pkg-config
     set(BLA_STATIC OFF)
     set(BLA_VENDOR "OpenBLAS")

cmake/qhotkey.cmake

@@ -15,10 +15,10 @@ ExternalProject_Add(qhotkey
         -DCMAKE_INSTALL_LIBDIR=lib
         -DCMAKE_POSITION_INDEPENDENT_CODE=ON
         -DCMAKE_POLICY_VERSION_MINIMUM=3.5
+        -DQT_DEFAULT_MAJOR_VERSION=6
     BUILD_ALWAYS False
 )
 
-find_package(Qt5 COMPONENTS X11Extras REQUIRED)
-
-list(APPEND deps_libs Qt5::X11Extras "${external_lib_dir}/libqhotkey.a")
+# Qt6: X11Extras removed, functionality now in QtGui via QNativeInterface
+list(APPEND deps_libs "${external_lib_dir}/libqhotkey.a")
 list(APPEND deps qhotkey)

cmake/translations.cmake

@@ -3,7 +3,7 @@ set(enabled_translations ar ca_ES cs de en es fr fr_CA it nl no pl ru sv sl tr_T
 # finished translations
 set(enabled_translations ar ca_ES de en es fr fr_CA it nl no pl ru sv sl tr_TR uk zh_CN zh_TW)
 
-find_package(Qt5 COMPONENTS Core LinguistTools)
+find_package(Qt6 COMPONENTS Core LinguistTools)
 
 set(ts_files "")
 foreach(lang ${enabled_translations})
@@ -24,6 +24,6 @@ function(ADD_TRANSLATIONS_RESOURCE res_file)
     set(${res_file} ${_res_file} PARENT_SCOPE)
 endfunction()
 
-qt5_create_translation(qm_files ${CMAKE_SOURCE_DIR}/src ${desktop_dir}/qml ${sfos_dir}/qml ${ts_files})
+qt_add_translation(qm_files ${ts_files})
 
 add_translations_resource(translations_res ${qm_files})

config/models.json

@@ -35916,14 +35916,37 @@
             "engine": "stt_whisper",
             "lang_id": "yo"
         },
-        {
-            "name": "中文 (FasterWhisper)",
-            "id": "zh_fasterwhisper",
-            "engine": "stt_fasterwhisper",
-            "lang_id": "zh"
-        },
-        {
-            "name": "中文 (WhisperCpp)",
+        {
+            "name": "中文 (FasterWhisper)",
+            "id": "zh_fasterwhisper",
+            "engine": "stt_fasterwhisper",
+            "lang_id": "zh"
+        },
+        {
+            "name": "Multilingual (Canary 1B v2)",
+            "model_id": "multilang_canary_1b_v2",
+            "engine": "stt_canary",
+            "lang_id": "multilang",
+            "info": "NVIDIA Canary 1B v2 - 4.89% WER, 5x faster than Whisper (RTFx 630), best accuracy-per-watt",
+            "options": "ti",
+            "score": 5,
+            "features": ["high_quality", "medium_processing", "stt_punctuation"],
+            "default_for_lang": true,
+            "hidden": false
+        },
+        {
+            "name": "Multilingual (Canary Qwen 2.5B)",
+            "model_id": "multilang_canary_qwen",
+            "engine": "stt_canary",
+            "lang_id": "multilang",
+            "info": "NVIDIA Canary Qwen 2.5B - Larger model for maximum accuracy",
+            "options": "ti",
+            "score": 4,
+            "features": ["high_quality", "slow_processing", "stt_punctuation"],
+            "hidden": false
+        },
+        {
+            "name": "中文 (WhisperCpp)",
             "id": "zh_whisper",
             "engine": "stt_whisper",
             "lang_id": "zh"

src/app_server.cpp

@@ -134,7 +134,7 @@ int app_server::request_another_instance(const cmd::options &options) {
                     auto model = qdbus_cast<QVariantMap>(
                         m.template value<QDBusArgument>());
                     return std::max(model.contains("id")
-                                        ? model.value("id").toString().size()
+                                        ? static_cast<int>(model.value("id").toString().size())
                                         : size,
                                     size);
                 });
@@ -204,25 +204,25 @@ int app_server::request_another_instance(const cmd::options &options) {
             g_max_size =
                 std::max(g_max_size,
                          std::max(modelStt.contains("id")
-                                      ? modelStt.value("id").toString().size()
+                                      ? static_cast<int>(modelStt.value("id").toString().size())
                                       : 1,
                                   modelTts.contains("id")
-                                      ? modelTts.value("id").toString().size()
+                                      ? static_cast<int>(modelTts.value("id").toString().size())
                                       : 1));
             print_active_model("STT", g_max_size, modelStt);
             print_active_model("TTS", g_max_size, modelTts);
         } else if (options.active_model_to_print_role & cmd::role_stt) {
             auto modelStt = iface.activeSttModel();
             g_max_size = std::max(g_max_size,
                                   modelStt.contains("id")
-                                      ? modelStt.value("id").toString().size()
+                                      ? static_cast<int>(modelStt.value("id").toString().size())
                                       : 1);
             print_active_model("STT", g_max_size, modelStt);
         } else if (options.active_model_to_print_role & cmd::role_tts) {
             auto modelTts = iface.activeTtsModel();
             g_max_size = std::max(g_max_size,
                                   modelTts.contains("id")
-                                      ? modelTts.value("id").toString().size()
+                                      ? static_cast<int>(modelTts.value("id").toString().size())
                                       : 1);
             print_active_model("TTS", g_max_size, modelTts);
         }