Detect wake word model from index.json (#1211)

* detect wake word model from index.json

* update wait time before entering wifi configure mode

Author: 78

CMakeLists.txt

@@ -4,7 +4,7 @@
 # CMakeLists in this exact order for cmake to work correctly
 cmake_minimum_required(VERSION 3.16)
 
-set(PROJECT_VER "2.0.1")
+set(PROJECT_VER "2.0.2")
 
 # Add this line to disable the specific warning
 add_compile_options(-Wno-missing-field-initializers)

main/CMakeLists.txt

@@ -45,6 +45,17 @@ list(APPEND SOURCES ${BOARD_COMMON_SOURCES})
 list(APPEND INCLUDE_DIRS ${CMAKE_CURRENT_SOURCE_DIR}/boards/common)
 
 idf_build_get_property(build_components BUILD_COMPONENTS)
+# Function to find component dynamically by pattern
+function(find_component_by_pattern PATTERN COMPONENT_VAR PATH_VAR)
+    foreach(COMPONENT ${build_components})
+        if(COMPONENT MATCHES "${PATTERN}")
+            set(${COMPONENT_VAR} ${COMPONENT} PARENT_SCOPE)
+            idf_component_get_property(COMPONENT_PATH ${COMPONENT} COMPONENT_DIR)
+            set(${PATH_VAR} "${COMPONENT_PATH}" PARENT_SCOPE)
+            break()
+        endif()
+    endforeach()
+endfunction()
 
 # Set default BUILTIN_TEXT_FONT and BUILTIN_ICON_FONT
 set(BUILTIN_TEXT_FONT font_puhui_14_1)
@@ -64,7 +75,7 @@ elseif(CONFIG_BOARD_TYPE_BREAD_COMPACT_ESP32)
     set(BOARD_TYPE "bread-compact-esp32")
 elseif(CONFIG_BOARD_TYPE_BREAD_COMPACT_ESP32_LCD)
     set(BOARD_TYPE "bread-compact-esp32-lcd")
-    set(BUILTIN_TEXT_FONT font_puhui_14_1)
+    set(BUILTIN_TEXT_FONT font_puhui_basic_14_1)
     set(BUILTIN_ICON_FONT font_awesome_14_1)
 elseif(CONFIG_BOARD_TYPE_DF_K10)
     set(BOARD_TYPE "df-k10")
@@ -202,14 +213,10 @@ elseif(CONFIG_BOARD_TYPE_ECHOEAR)
     set(BUILTIN_TEXT_FONT font_puhui_20_4)
     set(BUILTIN_ICON_FONT font_awesome_20_4)
     # Find esp_emote_gfx component for ECHOEAR extra files
-    foreach(COMPONENT ${build_components})
-        if(COMPONENT MATCHES "esp_emote_gfx" OR COMPONENT MATCHES "espressif2022__esp_emote_gfx")
-            set(EMOTE_GFX_COMPONENT ${COMPONENT})
-            idf_component_get_property(EMOTE_GFX_COMPONENT_PATH ${EMOTE_GFX_COMPONENT} COMPONENT_DIR)
-            set(DEFAULT_ASSETS_EXTRA_FILES "${EMOTE_GFX_COMPONENT_PATH}/emoji_normal")
-            break()
-        endif()
-    endforeach()
+    find_component_by_pattern("esp_emote_gfx" EMOTE_GFX_COMPONENT EMOTE_GFX_COMPONENT_PATH)
+    if(EMOTE_GFX_COMPONENT_PATH)
+        set(DEFAULT_ASSETS_EXTRA_FILES "${EMOTE_GFX_COMPONENT_PATH}/emoji_normal")
+    endif()
 elseif(CONFIG_BOARD_TYPE_ESP32S3_AUDIO_BOARD)
     set(BOARD_TYPE "waveshare-s3-audio-board")
     set(BUILTIN_TEXT_FONT font_puhui_basic_16_4)
@@ -406,11 +413,11 @@ elseif(CONFIG_BOARD_TYPE_GENJUTECH_S3_1_54TFT)
     set(DEFAULT_EMOJI_COLLECTION twemoji_64)
 elseif(CONFIG_BOARD_TYPE_ESP32_CGC)
     set(BOARD_TYPE "esp32-cgc")
-    set(BUILTIN_TEXT_FONT font_puhui_14_1)
+    set(BUILTIN_TEXT_FONT font_puhui_basic_14_1)
     set(BUILTIN_ICON_FONT font_awesome_14_1)
 elseif(CONFIG_BOARD_TYPE_ESP32_CGC_144)
     set(BOARD_TYPE "esp32-cgc-144")
-    set(BUILTIN_TEXT_FONT font_puhui_14_1)
+    set(BUILTIN_TEXT_FONT font_puhui_basic_14_1)
     set(BUILTIN_ICON_FONT font_awesome_14_1)
 elseif(CONFIG_BOARD_TYPE_ESP_S3_LCD_EV_Board)
     set(BOARD_TYPE "esp-s3-lcd-ev-board")
@@ -499,12 +506,11 @@ if(CONFIG_USE_AUDIO_PROCESSOR)
 else()
     list(APPEND SOURCES "audio/processors/no_audio_processor.cc")
 endif()
-if(CONFIG_USE_AFE_WAKE_WORD)
+if(CONFIG_IDF_TARGET_ESP32S3 OR CONFIG_IDF_TARGET_ESP32P4)
     list(APPEND SOURCES "audio/wake_words/afe_wake_word.cc")
-elseif(CONFIG_USE_ESP_WAKE_WORD)
-    list(APPEND SOURCES "audio/wake_words/esp_wake_word.cc")
-elseif(CONFIG_USE_CUSTOM_WAKE_WORD)
     list(APPEND SOURCES "audio/wake_words/custom_wake_word.cc")
+else()
+    list(APPEND SOURCES "audio/wake_words/esp_wake_word.cc")
 endif()
 
 # Select language directory according to Kconfig
@@ -603,24 +609,16 @@ add_custom_target(lang_header ALL
 )
 
 # Find ESP-SR component dynamically
-foreach(COMPONENT ${build_components})
-    if(COMPONENT MATCHES "espressif__esp-sr")
-        set(ESP_SR_COMPONENT ${COMPONENT})
-        idf_component_get_property(ESP_SR_COMPONENT_PATH ${ESP_SR_COMPONENT} COMPONENT_DIR)
-        set(ESP_SR_MODEL_PATH "${ESP_SR_COMPONENT_PATH}/model")
-        break()
-    endif()
-endforeach()
+find_component_by_pattern("espressif__esp-sr" ESP_SR_COMPONENT ESP_SR_COMPONENT_PATH)
+if(ESP_SR_COMPONENT_PATH)
+    set(ESP_SR_MODEL_PATH "${ESP_SR_COMPONENT_PATH}/model")
+endif()
 
 # Find xiaozhi-fonts component dynamically
-foreach(COMPONENT ${build_components})
-    if(COMPONENT MATCHES "xiaozhi-fonts")
-        set(XIAOZHI_FONTS_COMPONENT ${COMPONENT})
-        idf_component_get_property(XIAOZHI_FONTS_COMPONENT_PATH ${XIAOZHI_FONTS_COMPONENT} COMPONENT_DIR)
-        set(XIAOZHI_FONTS_PATH "${XIAOZHI_FONTS_COMPONENT_PATH}")
-        break()
-    endif()
-endforeach()
+find_component_by_pattern("xiaozhi-fonts" XIAOZHI_FONTS_COMPONENT XIAOZHI_FONTS_COMPONENT_PATH)
+if(XIAOZHI_FONTS_COMPONENT_PATH)
+    set(XIAOZHI_FONTS_PATH "${XIAOZHI_FONTS_COMPONENT_PATH}")
+endif()
 
 if(CONFIG_BOARD_TYPE_ESP_HI)
 set(URL "https://github.com/espressif2022/image_player/raw/main/test_apps/test_8bit")

main/audio/audio_service.cc

@@ -8,12 +8,11 @@
 #include "processors/no_audio_processor.h"
 #endif
 
-#if CONFIG_USE_AFE_WAKE_WORD
+#if CONFIG_IDF_TARGET_ESP32S3 || CONFIG_IDF_TARGET_ESP32P4
 #include "wake_words/afe_wake_word.h"
-#elif CONFIG_USE_ESP_WAKE_WORD
-#include "wake_words/esp_wake_word.h"
-#elif CONFIG_USE_CUSTOM_WAKE_WORD
 #include "wake_words/custom_wake_word.h"
+#else
+#include "wake_words/esp_wake_word.h"
 #endif
 
 #define TAG "AudioService"
@@ -50,16 +49,6 @@ void AudioService::Initialize(AudioCodec* codec) {
     audio_processor_ = std::make_unique<NoAudioProcessor>();
 #endif
 
-#if CONFIG_USE_AFE_WAKE_WORD
-    wake_word_ = std::make_unique<AfeWakeWord>();
-#elif CONFIG_USE_ESP_WAKE_WORD
-    wake_word_ = std::make_unique<EspWakeWord>();
-#elif CONFIG_USE_CUSTOM_WAKE_WORD
-    wake_word_ = std::make_unique<CustomWakeWord>();
-#else
-    wake_word_ = nullptr;
-#endif
-
     audio_processor_->OnOutput([this](std::vector<int16_t>&& data) {
         PushTaskToEncodeQueue(kAudioTaskTypeEncodeToSendQueue, std::move(data));
     });
@@ -71,14 +60,6 @@ void AudioService::Initialize(AudioCodec* codec) {
         }
     });
 
-    if (wake_word_) {
-        wake_word_->OnWakeWordDetected([this](const std::string& wake_word) {
-            if (callbacks_.on_wake_word_detected) {
-                callbacks_.on_wake_word_detected(wake_word);
-            }
-        });
-    }
-
     esp_timer_create_args_t audio_power_timer_args = {
         .callback = [](void* arg) {
             AudioService* audio_service = (AudioService*)arg;
@@ -100,11 +81,11 @@ void AudioService::Start() {
 
 #if CONFIG_USE_AUDIO_PROCESSOR
     /* Start the audio input task */
-    xTaskCreate([](void* arg) {
+    xTaskCreatePinnedToCore([](void* arg) {
         AudioService* audio_service = (AudioService*)arg;
         audio_service->AudioInputTask();
         vTaskDelete(NULL);
-    }, "audio_input", 2048 * 3, this, 8, &audio_input_task_handle_);
+    }, "audio_input", 2048 * 3, this, 8, &audio_input_task_handle_, 0);
 
     /* Start the audio output task */
     xTaskCreate([](void* arg) {
@@ -670,4 +651,36 @@ void AudioService::CheckAndUpdateAudioPowerState() {
 
 void AudioService::SetModelsList(srmodel_list_t* models_list) {
     models_list_ = models_list;
-}
+
+#if CONFIG_IDF_TARGET_ESP32S3 || CONFIG_IDF_TARGET_ESP32P4
+    if (esp_srmodel_filter(models_list_, ESP_MN_PREFIX, NULL) != nullptr) {
+        wake_word_ = std::make_unique<CustomWakeWord>();
+    } else if (esp_srmodel_filter(models_list_, ESP_WN_PREFIX, NULL) != nullptr) {
+        wake_word_ = std::make_unique<AfeWakeWord>();
+    } else {
+        wake_word_ = nullptr;
+    }
+#else
+    if (esp_srmodel_filter(models_list_, ESP_WN_PREFIX, NULL) != nullptr) {
+        wake_word_ = std::make_unique<EspWakeWord>();
+    } else {
+        wake_word_ = nullptr;
+    }
+#endif
+
+    if (wake_word_) {
+        wake_word_->OnWakeWordDetected([this](const std::string& wake_word) {
+            if (callbacks_.on_wake_word_detected) {
+                callbacks_.on_wake_word_detected(wake_word);
+            }
+        });
+    }
+}
+
+bool AudioService::IsAfeWakeWord() {
+#if CONFIG_IDF_TARGET_ESP32S3 || CONFIG_IDF_TARGET_ESP32P4
+    return wake_word_ != nullptr && dynamic_cast<AfeWakeWord*>(wake_word_.get()) != nullptr;
+#else
+    return false;
+#endif
+}

main/audio/audio_service.h

@@ -94,6 +94,7 @@ class AudioService {
     bool IsIdle();
     bool IsWakeWordRunning() const { return xEventGroupGetBits(event_group_) & AS_EVENT_WAKE_WORD_RUNNING; }
     bool IsAudioProcessorRunning() const { return xEventGroupGetBits(event_group_) & AS_EVENT_AUDIO_PROCESSOR_RUNNING; }
+    bool IsAfeWakeWord();
 
     void EnableWakeWordDetection(bool enable);
     void EnableVoiceProcessing(bool enable);

main/audio/wake_words/custom_wake_word.cc

@@ -91,8 +91,10 @@ bool CustomWakeWord::Initialize(AudioCodec* codec, srmodel_list_t* models_list)
     if (models_list == nullptr) {
         language_ = "cn";
         models_ = esp_srmodel_init("model");
+#if CONFIG_CUSTOM_WAKE_WORD
         threshold_ = CONFIG_CUSTOM_WAKE_WORD_THRESHOLD / 100.0f;
         commands_.push_back({CONFIG_CUSTOM_WAKE_WORD, CONFIG_CUSTOM_WAKE_WORD_DISPLAY, "wake"});
+#endif
     } else {
         models_ = models_list;
         ParseWakenetModelConfig();

main/boards/atommatrix-echo-base/README.md

@@ -18,18 +18,6 @@ idf.py menuconfig
 Xiaozhi Assistant -> Board Type -> AtomMatrix + Echo Base
 ```
 
-**修改 flash 大小：**
-
-```
-Serial flasher config -> Flash size -> 4 MB
-```
-
-**修改分区表：**
-
-```
-Partition Table -> Custom partition CSV file -> partitions/v1/4m.csv
-```
-
 **编译：**
 
 ```bash

main/boards/atommatrix-echo-base/config.json

@@ -4,8 +4,6 @@
         {
             "name": "atommatrix-echo-base",
             "sdkconfig_append": [
-                "CONFIG_ESPTOOLPY_FLASHSIZE_4MB=y",
-                "CONFIG_PARTITION_TABLE_CUSTOM_FILENAME=\"partitions/v1/4m.csv\""
             ]
         }
     ]

main/boards/atoms3-echo-base/config.json

@@ -5,7 +5,6 @@
             "name": "atoms3-echo-base",
             "sdkconfig_append": [
                 "CONFIG_SPIRAM=n",
-                "CONFIG_USE_AFE=n",
                 "CONFIG_ESPTOOLPY_FLASHSIZE_8MB=y",
                 "CONFIG_PARTITION_TABLE_CUSTOM_FILENAME=\"partitions/v2/8m.csv\""
             ]

main/boards/bread-compact-esp32-lcd/config.json

@@ -4,8 +4,6 @@
         {
             "name": "bread-compact-esp32-lcd",
             "sdkconfig_append": [
-                "CONFIG_ESPTOOLPY_FLASHSIZE_4MB=y",
-                "CONFIG_PARTITION_TABLE_CUSTOM_FILENAME=\"partitions/v1/4m.csv\"",
                 "LCD_ST7789_240X240_7PIN=y"
             ]
         }

main/boards/bread-compact-esp32/README.md

@@ -18,18 +18,6 @@ idf.py menuconfig
 Xiaozhi Assistant -> Board Type -> 面包板 ESP32 DevKit
 ```
 
-**修改 flash 大小：**
-
-```
-Serial flasher config -> Flash size -> 4 MB
-```
-
-**修改分区表：**
-
-```
-Partition Table -> Custom partition CSV file -> partitions/v1/4m.csv
-```
-
 **编译：**
 
 ```bash