feat(voice-button): refactor speech recognition and transcript insertion logic

Author: SonyLeo

packages/components/src/sender-actions/voice-button/index.vue

@@ -34,31 +34,31 @@ const insertTranscript = (transcript: string) => {
     return
   }
 
-  // 在单次录音会话期间，持续替换当前的语音插入范围
+  // autoReplace 模式：替换当前范围内的内容
   const range = speechRange.value ?? {
     from: editorInstance.state.selection.from,
     to: editorInstance.state.selection.to,
   }
-  const tr = editorInstance.state.tr.insertText(transcript, range.from, range.to)
-  editorInstance.view.dispatch(tr)
+
+  // 用 insertContentAt 传 range，会替换该范围内的内容
+  editorInstance.commands.insertContentAt(range, transcript)
+
+  // 更新范围，下次调用时覆盖本次插入的内容
   speechRange.value = {
     from: range.from,
     to: range.from + transcript.length,
   }
+
   editorInstance.commands.focus('end')
 }
 
-// 语音配置 - 使用普通对象而不是 computed，避免每次都创建新对象
 const speechOptions = {
   ...props.speechConfig,
   onStart: () => {
     resetSpeechRange()
     emit('speech-start')
   },
   onInterim: (transcript: string) => {
-    if (props.speechConfig?.autoReplace) {
-      insertTranscript(transcript)
-    }
     emit('speech-interim', transcript)
   },
   onFinal: (transcript: string) => {

packages/components/src/sender-actions/voice-button/webSpeechHandler.ts

@@ -52,14 +52,25 @@ export class WebSpeechHandler implements SpeechHandler {
       callbacks.onEnd()
     }
     this.recognition.onresult = (event: SpeechRecognitionEvent) => {
-      const transcript = Array.from(event.results)
-        .map((result) => result[0].transcript)
-        .join('')
-      const current = event.results[event.resultIndex]
-      if (current?.isFinal) {
-        callbacks.onFinal(transcript)
-      } else {
-        callbacks.onInterim(transcript)
+      // 标准做法：分离 interim 和 final 结果
+      let interimTranscript = ''
+      let finalTranscript = ''
+
+      // 只处理 resultIndex 之后的新增 results
+      for (let i = event.resultIndex; i < event.results.length; i++) {
+        const transcript = event.results[i][0].transcript
+        if (event.results[i].isFinal) {
+          finalTranscript += transcript
+        } else {
+          interimTranscript += transcript
+        }
+      }
+
+      // 如果有 final 结果，优先触发 onFinal
+      if (finalTranscript) {
+        callbacks.onFinal(finalTranscript)
+      } else if (interimTranscript) {
+        callbacks.onInterim(interimTranscript)
       }
     }
     this.recognition.onerror = (event: SpeechRecognitionErrorEvent) => {