diff --git a/workflow/service/flow/lambda_node_imp.go b/workflow/service/flow/lambda_node_imp.go
index 8019385..972310f 100644
--- a/workflow/service/flow/lambda_node_imp.go
+++ b/workflow/service/flow/lambda_node_imp.go
@@ -683,24 +683,29 @@ func AudioOptimizeNode(ctx context.Context, nodeInput *flowDto.NodeExecutionInpu
 	return outputRes, nil
 }
 
+// splitTextByPunct 按中文标点分割句子，同时保留标点在分段内
+// 例如："这个叫高血压调理方，注意是根源调理不是临时缓解，"
+// 会变成：["这个叫高血压调理方，", "注意是根源调理不是临时缓解，"]
 func splitTextByPunct(raw string) []string {
-	// 按标点切分+拼接标点
-	slice := regexp.MustCompile(`([，。；！？])`).Split(raw, -1)
-	var res []string
-	var builder strings.Builder
-	for idx, s := range slice {
-		if s == "" {
-			continue
-		}
-		builder.WriteString(s)
-		// 偶数位是分隔标点（split后规律：文本、标点、文本、标点...）
-		if idx%2 == 1 {
-			res = append(res, builder.String())
-			builder.Reset()
-		}
+	// 匹配中文标点并保留在文本中，按标点位置切分
+	re := regexp.MustCompile(`[，。；！？]`)
+	// 先找到所有标点的位置
+	indexes := re.FindAllStringIndex(raw, -1)
+	if len(indexes) == 0 {
+		return []string{raw}
 	}
-	if builder.Len() > 0 {
-		res = append(res, builder.String())
+
+	var res []string
+	prev := 0
+	for _, idx := range indexes {
+		end := idx[1] // 标点的结束位置
+		seg := raw[prev:end]
+		res = append(res, seg)
+		prev = end
+	}
+	// 处理最后一段没有标点的文本
+	if prev < len(raw) {
+		res = append(res, raw[prev:])
 	}
 	return res
 }
@@ -708,46 +713,53 @@ func splitTextByPunct(raw string) []string {
 // BuildSubtitles 核心工具：单个sentence生成多条subtitle
 func BuildSubtitles(sents *[]flowDto.Sentence) ([]flowDto.Subtitle, error) {
 	var subtitles []flowDto.Subtitle
+
 	for _, sent := range *sents {
+		// 1. 先按标点把文本拆成多个片段（保留标点）
 		segList := splitTextByPunct(sent.Text)
 		if len(segList) == 0 {
-			return nil, nil
+			continue
 		}
 
-		var subs []flowDto.Subtitle
 		wordIdx := 0
 		allWords := sent.Words
-
+		// 2. 遍历每个文本片段，匹配对应的Words
 		for _, seg := range segList {
+			// 去除文本片段的标点，方便和Word.Word拼接内容匹配
+			segClean := strings.ReplaceAll(seg, "，", "")
+			segClean = strings.ReplaceAll(segClean, "。", "")
+			segClean = strings.ReplaceAll(segClean, "；", "")
+			segClean = strings.ReplaceAll(segClean, "！", "")
+			segClean = strings.ReplaceAll(segClean, "？", "")
+
 			var collectWords []flowDto.Word
-			currentText := ""
-			// 循环取 word，直到拼接内容 包含/匹配 seg
-			for {
-				if wordIdx >= len(allWords) {
-					break
-				}
+			var currentText strings.Builder
+
+			// 收集Word直到拼接内容覆盖当前分段
+			for wordIdx < len(allWords) {
 				word := allWords[wordIdx]
-				currentText += word.Word
+				currentText.WriteString(word.Word)
 				collectWords = append(collectWords, word)
 				wordIdx++
 
-				// 只要包含分段文本，就认为匹配（无视末尾标点差异）
-				if strings.Contains(currentText, seg) {
+				// 当拼接的文本包含当前分段的纯文本时，停止收集
+				if strings.Contains(currentText.String(), segClean) {
 					break
 				}
 			}
+
 			if len(collectWords) == 0 {
 				continue
 			}
-			// 生成字幕
+
+			// 3. 生成字幕（时间戳取首尾Word的时间）
 			sub := flowDto.Subtitle{
 				Start: collectWords[0].StartTime,
 				End:   collectWords[len(collectWords)-1].EndTime,
-				Text:  seg,
+				Text:  segClean,
 			}
-			subs = append(subs, sub)
+			subtitles = append(subtitles, sub)
 		}
-		subtitles = append(subtitles, subs...)
 	}
 
 	return subtitles, nil
@@ -792,12 +804,12 @@ func DataConversionNode(ctx context.Context, nodeInput *flowDto.NodeExecutionInp
 	}
 
 	jsonStr := ``
-	jsonVal := "输出字段规范："
+	jsonVal := ""
 	for _, field := range nodeInput.Config.OutputConfig {
 		jsonStr, _ = sjson.Set(jsonStr, field.Field, "")
-		jsonVal += fmt.Sprintf("%s:%s;", field.Field, field.Value)
+		//jsonVal += fmt.Sprintf("%s:%s;", field.Field, field.Value)
 	}
-	jsonVal += fmt.Sprintf("输出模板结构,仅修改每个字段对应数值：%s", jsonStr)
+	jsonVal += fmt.Sprintf("输出字段规范：%v", jsonStr)
 	nodeInput.Config.PromptContent = fmt.Sprintf("%s;%s", nodeInput.Config.PromptContent, jsonVal)
 	mapTaskResult, err := GetModelResult(ctx, "", nodeInput, skillName, form, userForm)
 	if err != nil {