diff --git a/internal/tools/youtube/timestamp_test.go b/internal/tools/youtube/timestamp_test.go
new file mode 100644
index 00000000..34f23c6f
--- /dev/null
+++ b/internal/tools/youtube/timestamp_test.go
@@ -0,0 +1,60 @@
+package youtube
+
+import (
+	"testing"
+)
+
+func TestParseTimestampToSeconds(t *testing.T) {
+	tests := []struct {
+		timestamp string
+		expected  int
+		shouldErr bool
+	}{
+		{"00:30", 30, false},
+		{"01:30", 90, false},
+		{"01:05:30", 3930, false}, // 1 hour 5 minutes 30 seconds
+		{"10:00", 600, false},
+		{"invalid", 0, true},
+		{"1:2:3:4", 0, true}, // too many parts
+	}
+
+	for _, test := range tests {
+		result, err := parseTimestampToSeconds(test.timestamp)
+
+		if test.shouldErr {
+			if err == nil {
+				t.Errorf("Expected error for timestamp %s, but got none", test.timestamp)
+			}
+		} else {
+			if err != nil {
+				t.Errorf("Unexpected error for timestamp %s: %v", test.timestamp, err)
+			}
+			if result != test.expected {
+				t.Errorf("For timestamp %s, expected %d seconds, got %d", test.timestamp, test.expected, result)
+			}
+		}
+	}
+}
+
+func TestShouldIncludeRepeat(t *testing.T) {
+	tests := []struct {
+		lastTimestamp    string
+		currentTimestamp string
+		expected         bool
+		description      string
+	}{
+		{"00:30", "01:30", true, "60 second gap should allow repeat"},
+		{"00:30", "00:45", false, "15 second gap should not allow repeat"},
+		{"01:00", "01:30", true, "30 second gap should allow repeat (boundary case)"},
+		{"01:00", "01:29", false, "29 second gap should not allow repeat"},
+		{"invalid", "01:30", true, "invalid timestamp should err on side of inclusion"},
+		{"01:30", "invalid", true, "invalid timestamp should err on side of inclusion"},
+	}
+
+	for _, test := range tests {
+		result := shouldIncludeRepeat(test.lastTimestamp, test.currentTimestamp)
+		if result != test.expected {
+			t.Errorf("%s: expected %v, got %v", test.description, test.expected, result)
+		}
+	}
+}
diff --git a/internal/tools/youtube/youtube.go b/internal/tools/youtube/youtube.go
index f5bac86f..ac8a5cf3 100644
--- a/internal/tools/youtube/youtube.go
+++ b/internal/tools/youtube/youtube.go
@@ -30,7 +30,9 @@ import (
 )
 
 // Match timestamps like "00:00:01.234" or just numbers or sequence numbers
-var timestampRegex = regexp.MustCompile(`^\d+$|^\d{1,2}:\d{2}(:\d{2})?(\\.\d{3})?$`)
+var timestampRegex = regexp.MustCompile(`^\d+$|^\d{1,2}:\d{2}(:\d{2})?(\.\d{3})?$`)
+
+const TimeGapForRepeats = 10 // seconds
 
 func NewYouTube() (ret *YouTube) {
 
@@ -222,7 +224,10 @@ func (o *YouTube) readAndFormatVTTWithTimestamps(filename string) (ret string, e
 	lines := strings.Split(string(content), "\n")
 	var textBuilder strings.Builder
 	var currentTimestamp string
-	seenSegments := make(map[string]struct{})
+	// Track content with timestamps to allow repeats after significant time gaps
+	// This preserves legitimate repeated content (choruses, recurring phrases, etc.)
+	// while still filtering out immediate duplicates from VTT formatting issues
+	seenSegments := make(map[string]string) // text -> last timestamp seen
 
 	for _, line := range lines {
 		line = strings.TrimSpace(line)
@@ -254,11 +259,19 @@ func (o *YouTube) readAndFormatVTTWithTimestamps(filename string) (ret string, e
 			// Remove VTT formatting tags
 			cleanText := removeVTTTags(line)
 			if cleanText != "" && currentTimestamp != "" {
-				// Use just the clean text as the key to avoid duplicates across different timestamps
-				if _, exists := seenSegments[cleanText]; !exists {
+				// Check if we should include this segment
+				shouldInclude := true
+				if lastTimestamp, exists := seenSegments[cleanText]; exists {
+					// Calculate time difference to determine if this is a legitimate repeat
+					if !shouldIncludeRepeat(lastTimestamp, currentTimestamp) {
+						shouldInclude = false
+					}
+				}
+
+				if shouldInclude {
 					timestampedLine := fmt.Sprintf("[%s] %s", currentTimestamp, cleanText)
 					textBuilder.WriteString(timestampedLine + "\n")
-					seenSegments[cleanText] = struct{}{}
+					seenSegments[cleanText] = currentTimestamp
 				}
 			}
 		}
@@ -290,6 +303,59 @@ func removeVTTTags(s string) string {
 	return tagRegex.ReplaceAllString(s, "")
 }
 
+// shouldIncludeRepeat determines if repeated content should be included based on time gap
+func shouldIncludeRepeat(lastTimestamp, currentTimestamp string) bool {
+	// Parse timestamps to calculate time difference
+	lastSeconds, err1 := parseTimestampToSeconds(lastTimestamp)
+	currentSeconds, err2 := parseTimestampToSeconds(currentTimestamp)
+
+	if err1 != nil || err2 != nil {
+		// If we can't parse timestamps, err on the side of inclusion
+		return true
+	}
+
+	// Allow repeats if there's at least a TimeGapForRepeats gap
+	// This threshold can be adjusted based on use case:
+	// - 10 seconds works well for most content
+	// - Could be made configurable in the future
+	timeDiffSeconds := currentSeconds - lastSeconds
+	return timeDiffSeconds >= TimeGapForRepeats
+}
+
+// parseTimestampToSeconds converts timestamp string (HH:MM:SS or MM:SS) to total seconds
+func parseTimestampToSeconds(timestamp string) (int, error) {
+	parts := strings.Split(timestamp, ":")
+	if len(parts) < 2 || len(parts) > 3 {
+		return 0, fmt.Errorf("invalid timestamp format: %s", timestamp)
+	}
+
+	var hours, minutes, seconds int
+	var err error
+
+	if len(parts) == 3 {
+		// HH:MM:SS format
+		if hours, err = strconv.Atoi(parts[0]); err != nil {
+			return 0, err
+		}
+		if minutes, err = strconv.Atoi(parts[1]); err != nil {
+			return 0, err
+		}
+		if seconds, err = strconv.Atoi(parts[2]); err != nil {
+			return 0, err
+		}
+	} else {
+		// MM:SS format
+		if minutes, err = strconv.Atoi(parts[0]); err != nil {
+			return 0, err
+		}
+		if seconds, err = strconv.Atoi(parts[1]); err != nil {
+			return 0, err
+		}
+	}
+
+	return hours*3600 + minutes*60 + seconds, nil
+}
+
 func (o *YouTube) GrabComments(videoId string) (ret []string, err error) {
 	if err = o.initService(); err != nil {
 		return