feat: 增加了音频输入选项，并优化了字幕引擎的构建和运行流程。

- 新增了系统音频输入（麦克风）的选择功能
- 重构了字幕引擎的构建流程，使用 PyInstaller 打包为可执行文件
- 优化了字幕引擎的启动和停止逻辑
- 更新了用户界面，增加了音频选择的控制选项
- 修改了相关的文件路径和构建配置

src/main/index.ts

@@ -2,6 +2,7 @@ import { app, BrowserWindow } from 'electron'
 import { electronApp, optimizer } from '@electron-toolkit/utils'
 import { controlWindow } from './control'
 import { captionWindow } from './caption'
+import { captionEngine } from './utils/config'
 
 app.whenReady().then(() => {
   electronApp.setAppUserModelId('com.himeditator.autocaption')
@@ -22,6 +23,10 @@ app.whenReady().then(() => {
   })
 })
 
+app.on('will-quit', async () => { 
+  captionEngine.stop()
+});
+
 app.on('window-all-closed', () => {
   if (process.platform !== 'darwin') {
     app.quit()

src/main/types/index.ts

@@ -23,6 +23,7 @@ export interface Controls {
   sourceLang: string,
   targetLang: string,
   engine: string,
+  audio: 0 | 1,
   translation: boolean,
   customized: boolean,
   customizedApp: string,

src/main/utils/config.ts

@@ -22,6 +22,7 @@ export const controls: Controls = {
   sourceLang: 'en',
   targetLang: 'zh',
   engine: 'gummy',
+  audio: 0,
   engineEnabled: false,
   translation: true,
   customized: false,
@@ -74,6 +75,7 @@ export function setControls(args: any) {
   controls.sourceLang = args.sourceLang
   controls.targetLang = args.targetLang
   controls.engine = args.engine
+  controls.audio = args.audio
   controls.translation = args.translation
   controls.customized = args.customized
   controls.customizedApp = args.customizedApp

src/main/utils/engine.ts

@@ -1,5 +1,6 @@
-import { spawn } from 'child_process'
+import { spawn, exec } from 'child_process'
 import { app } from 'electron'
+import { is } from '@electron-toolkit/utils'
 import path from 'path'
 import { addCaptionLog, controls } from './config'
 
@@ -14,24 +15,29 @@ export class CaptionEngine {
             this.command = [ controls.customizedCommand ]
         }
         else if(controls.engine === 'gummy'){
-            this.appPath = path.join(
-                app.getAppPath(),
-                'python-subprocess', 'subenv', 'Scripts', 'python.exe'
-            )
+            if(is.dev){
+                this.appPath = path.join(
+                    app.getAppPath(),
+                    'python-subprocess', 'dist', 'main-gummy.exe'
+                )
+            }
+            else{
+                this.appPath = path.join(
+                    process.resourcesPath,
+                    'python-subprocess', 'dist', 'main-gummy.exe'
+                )
+            }
             this.command = []
-            this.command.push(path.join(
-                app.getAppPath(),
-                'python-subprocess', 'main.py'
-            ))
             this.command.push('-s', controls.sourceLang)
             this.command.push('-t',  controls.translation ? controls.targetLang : 'none')
+            this.command.push('-a', controls.audio ? '1' : '0')
 
-            console.log(this.appPath)
-            console.log(this.command)
+            console.log('[INFO] engine', this.appPath)
+            console.log('[INFO] engine command',this.command)
         }
     }
 
-    public start() { 
+    public start() {
         if (this.process) {
             this.stop();
         }
@@ -70,7 +76,15 @@ export class CaptionEngine {
 
     public stop() {
         if (this.process) {
-            this.process.kill();
+            if (process.platform === "win32" && this.process.pid) {
+                exec(`taskkill /pid ${this.process.pid} /t /f`, (error) => {
+                    if (error) {
+                        console.error(`Failed to kill process: ${error}`);
+                    }
+                });
+            } else {
+                this.process.kill('SIGKILL');
+            }
             this.process = undefined;
             controls.engineEnabled = false;
             console.log('[INFO] Caption engine process stopped');

src/renderer/src/components/CaptionControl.vue

@@ -29,6 +29,14 @@
         :options="captionEngine"
       ></a-select>
     </div>
+    <div class="control-item">
+      <span class="control-label">音频选择</span>
+      <a-select
+        class="control-input"
+        v-model:value="currentAudio"
+        :options="audioType"
+      ></a-select>
+    </div>
     <div class="control-item">
       <span class="control-label">启用翻译</span>
       <a-switch v-model:checked="currentTranslation" />
@@ -62,13 +70,15 @@
 import { ref, computed, watch } from 'vue'
 import { storeToRefs } from 'pinia'
 import { useCaptionControlStore } from '@renderer/stores/captionControl'
+import { notification } from 'ant-design-vue'
 
 const captionControl = useCaptionControlStore()
-const { captionEngine, changeSignal } = storeToRefs(captionControl)
+const { captionEngine, audioType, changeSignal } = storeToRefs(captionControl)
 
 const currentSourceLang = ref('auto')
 const currentTargetLang = ref('zh')
 const currentEngine = ref('gummy')
+const currentAudio = ref<0 | 1>(0)
 const currentTranslation = ref<boolean>(false)
 
 const currentCustomized = ref<boolean>(false)
@@ -88,6 +98,7 @@ function applyChange(){
   captionControl.sourceLang = currentSourceLang.value
   captionControl.targetLang = currentTargetLang.value
   captionControl.engine = currentEngine.value
+  captionControl.audio = currentAudio.value
   captionControl.translation = currentTranslation.value
 
   captionControl.customized = currentCustomized.value
@@ -95,12 +106,18 @@ function applyChange(){
   captionControl.customizedCommand = currentCustomizedCommand.value
 
   captionControl.sendControlChange()
+
+  notification.open({
+      message: '字幕控制已更改',
+      description: '如果字幕引擎已经启动，需要关闭后重启才会生效'
+  });
 }
 
 function cancelChange(){
   currentSourceLang.value = captionControl.sourceLang
   currentTargetLang.value = captionControl.targetLang
   currentEngine.value = captionControl.engine
+  currentAudio.value = captionControl.audio
   currentTranslation.value = captionControl.translation
 
   currentCustomized.value = captionControl.customized

src/renderer/src/stores/captionControl.ts

@@ -16,11 +16,23 @@ export const useCaptionControlStore = defineStore('captionControl', () => {
       ]
     },
   ])
+  const audioType = ref([
+    {
+      value: 0,
+      label: '系统音频输出（扬声器）'
+    },
+    {
+      value: 1,
+      label: '系统音频输入（麦克风）'
+    }
+  ])
+
   const engineEnabled = ref(false)
 
   const sourceLang = ref<string>('en')
   const targetLang = ref<string>('zh')
   const engine = ref<string>('gummy')
+  const audio = ref<0 | 1>(0)
   const translation = ref<boolean>(true)
   const customized = ref<boolean>(false)
   const customizedApp = ref<string>('')
@@ -34,6 +46,7 @@ export const useCaptionControlStore = defineStore('captionControl', () => {
       sourceLang: sourceLang.value,
       targetLang: targetLang.value,
       engine: engine.value,
+      audio: audio.value,
       translation: translation.value,
       customized: customized.value,
       customizedApp: customizedApp.value,
@@ -54,6 +67,7 @@ export const useCaptionControlStore = defineStore('captionControl', () => {
     sourceLang.value = controls.sourceLang
     targetLang.value = controls.targetLang
     engine.value = controls.engine
+    audio.value = controls.audio
     translation.value = controls.translation
     customized.value = controls.customized
     customizedApp.value = controls.customizedApp
@@ -73,7 +87,8 @@ export const useCaptionControlStore = defineStore('captionControl', () => {
     engineEnabled.value = true
     notification.open({
       message: '字幕引擎启动',
-      description: `原语言：${sourceLang.value}，是否翻译：${translation.value?'是':'否'}` + 
+      description: `原语言：${sourceLang.value}，是否翻译：${translation.value?'是':'否'}，` + 
+        `字幕引擎：${engine.value}，音频类型：${audio.value ? '输入音频' : '输出音频'}` +
         (translation.value ? `，翻译语言：${targetLang.value}` : '')
     });
   })
@@ -88,10 +103,12 @@ export const useCaptionControlStore = defineStore('captionControl', () => {
 
   return {
     captionEngine,      // 字幕引擎
+    audioType,          // 音频类型
     engineEnabled,      // 字幕引擎是否启用
     sourceLang,         // 源语言
     targetLang,         // 目标语言
     engine,             // 字幕引擎
+    audio,              // 选择音频
     translation,        // 是否启用翻译
     customized,         // 是否使用自定义字幕引擎
     customizedApp,      // 自定义字幕引擎的应用程序