diff --git a/src/subtitle/vtt.js b/src/subtitle/vtt.js
index 1a20162..851905f 100644
--- a/src/subtitle/vtt.js
+++ b/src/subtitle/vtt.js
@@ -1,39 +1,94 @@
-function millisecondsStringToNumber(msString) {
-  const cleanString = msString.trim();
-  const milliseconds = parseInt(cleanString, 10);
+/**
+ * 将多种格式的VTT时间戳字符串转换为毫秒数。
+ * 兼容以下格式：
+ * - mmm (e.g., "291040")
+ * - MM:SS (e.g., "00:03")
+ * - HH:MM:SS (e.g., "01:02:03")
+ * - MM:SS.mmm (e.g., "00:07.980")
+ * - HH:MM:SS.mmm (e.g., "01:02:03.456")
+ * - MM:SS:mmm (e.g., "00:07:536")
+ *
+ * @param {string} timestamp - VTT时间戳字符串.
+ * @returns {number} - 转换后的总毫秒数.
+ */
+function parseTimestampToMilliseconds(timestamp) {
+  const ts = timestamp.trim();
 
-  if (isNaN(milliseconds)) {
-    return 0;
+  if (!ts.includes(":") && !ts.includes(".")) {
+    return parseInt(ts, 10) || 0;
   }
 
-  return milliseconds;
+  let timePart = ts;
+  let msPart = "0";
+
+  if (ts.includes(".")) {
+    const parts = ts.split(".");
+    timePart = parts[0];
+    msPart = parts[1];
+  } else {
+    const colonParts = ts.split(":");
+    if (
+      colonParts.length > 1 &&
+      colonParts[colonParts.length - 1].length === 3
+    ) {
+      msPart = colonParts.pop();
+      timePart = colonParts.join(":");
+    }
+  }
+
+  const timeComponents = timePart.split(":").map((p) => parseInt(p, 10) || 0);
+  let hours = 0,
+    minutes = 0,
+    seconds = 0;
+
+  if (timeComponents.length === 3) {
+    [hours, minutes, seconds] = timeComponents;
+  } else if (timeComponents.length === 2) {
+    [minutes, seconds] = timeComponents;
+  } else if (timeComponents.length === 1) {
+    [seconds] = timeComponents;
+  }
+
+  const milliseconds = parseInt(msPart.padEnd(3, "0"), 10) || 0;
+
+  return (hours * 3600 + minutes * 60 + seconds) * 1000 + milliseconds;
 }
 
+/**
+ * 解析包含双语字幕的VTT文件内容。
+ * @param {string} vttText - VTT文件的文本内容。
+ * @returns {Array<Object>} 一个包含字幕对象的数组，每个对象包含 start, end, text, 和 translation.
+ */
 export function parseBilingualVtt(vttText) {
   const cleanText = vttText.replace(/^\uFEFF/, "").trim();
-  const cues = cleanText.split(/\n\n+/);
+  if (!cleanText) {
+    return [];
+  }
 
+  const cues = cleanText.split(/\n\n+/);
   const result = [];
 
-  for (const cue of cues) {
+  const startIndex = cues[0].toUpperCase().includes("WEBVTT") ? 1 : 0;
+
+  for (let i = startIndex; i < cues.length; i++) {
+    const cue = cues[i];
     if (!cue.includes("-->")) continue;
 
     const lines = cue.split("\n");
-
     const timestampLineIndex = lines.findIndex((line) => line.includes("-->"));
     if (timestampLineIndex === -1) continue;
 
     const [startTimeString, endTimeString] =
-      lines[timestampLineIndex].split(" --> ");
+      lines[timestampLineIndex].split("-->");
     const textLines = lines.slice(timestampLineIndex + 1);
 
     if (startTimeString && endTimeString && textLines.length > 0) {
-      const originalText = textLines[0].trim();
-      const translatedText = (textLines[1] || "").trim();
+      const originalText = textLines[0]?.trim() || "";
+      const translatedText = textLines[1]?.trim() || "";
 
       result.push({
-        start: millisecondsStringToNumber(startTimeString),
-        end: millisecondsStringToNumber(endTimeString),
+        start: parseTimestampToMilliseconds(startTimeString),
+        end: parseTimestampToMilliseconds(endTimeString),
         text: originalText,
         translation: translatedText,
       });