fix(media): m4a/AAC audio files from non-Apple sources are misidentified as video/mp4 and excluded from audio workflows (#111177)

YangManBOBO · steipete · web-flow · commit e45ebac97abb · 2026-07-19T22:18:23.000-07:00
* fix(media): preserve audio hints for isom-brand m4a/AAC files misidentified as video/mp4

* fix(media): correct ambiguous M4A aliases

Co-authored-by: 潘晓波0668000512 &lt;pan.xiaobo@xydigit.com&gt;

---------

Co-authored-by: Peter Steinberger &lt;steipete@gmail.com&gt;
diff --git a/packages/media-core/src/mime.test.ts b/packages/media-core/src/mime.test.ts
@@ -26,6 +26,12 @@ async function makeOoxmlZip(opts: { mainMime: string; partPath: string }): Promi
   return await zip.generateAsync({ type: "nodebuffer" });
 }
 
+// file-type classifies this generic ISO-BMFF brand as video/mp4 without track metadata.
+const ISOM_BRAND_BUFFER = Buffer.from(
+  "0000001c6674797069736f6d0000000069736f6d0000000000000000",
+  "hex",
+);
+
 describe("mime detection", () => {
   async function expectDetectedMime(params: {
     input: Parameters<typeof detectMime>[0];
@@ -221,10 +227,51 @@ describe("mime detection", () => {
     });
   });
 
-  it("preserves audio metadata when an MP4 extension cannot identify track kind", async () => {
-    await expectDetectedMime({
-      input: { filePath: "voice.mp4", headerMime: "audio/mp4" },
+  it.each([
+    {
+      name: "audio/mp4 header",
+      filePath: "voice.mp4",
+      headerMime: "audio/mp4",
       expected: "audio/mp4",
+    },
+    {
+      name: "audio/x-m4a header",
+      filePath: "voice.m4a",
+      headerMime: "audio/x-m4a",
+      expected: "audio/x-m4a",
+    },
+    {
+      name: "audio/m4a header",
+      filePath: "voice.m4a",
+      headerMime: "audio/m4a",
+      expected: "audio/m4a",
+    },
+    {
+      name: "m4a extension",
+      filePath: "voice.m4a",
+      headerMime: undefined,
+      expected: "audio/x-m4a",
+    },
+    {
+      name: "mp4 extension without an audio hint",
+      filePath: "clip.mp4",
+      headerMime: undefined,
+      expected: "video/mp4",
+    },
+    {
+      name: "audio/aac elementary-stream metadata",
+      filePath: "voice.aac",
+      headerMime: "audio/aac",
+      expected: "video/mp4",
+    },
+  ] as const)("resolves ambiguous isom-brand bytes from $name", async (testCase) => {
+    await expectDetectedMime({
+      input: {
+        buffer: ISOM_BRAND_BUFFER,
+        filePath: testCase.filePath,
+        headerMime: testCase.headerMime,
+      },
+      expected: testCase.expected,
     });
   });
 
@@ -381,6 +428,7 @@ describe("extensionForMime", () => {
     { mime: "audio/x-wav", expected: ".wav" },
     { mime: "audio/webm", expected: ".webm" },
     { mime: "audio/x-m4a", expected: ".m4a" },
+    { mime: "audio/m4a", expected: ".m4a" },
     { mime: "audio/mp4", expected: ".m4a" },
     { mime: "video/x-msvideo", expected: ".avi" },
     { mime: "video/mp4", expected: ".mp4" },
diff --git a/packages/media-core/src/mime.ts b/packages/media-core/src/mime.ts
@@ -28,6 +28,7 @@ const EXT_BY_MIME: Record<string, string> = {
   "audio/opus": ".opus",
   "audio/webm": ".webm",
   "audio/x-m4a": ".m4a",
+  "audio/m4a": ".m4a",
   "audio/mp4": ".m4a",
   "audio/x-caf": ".caf",
   "video/x-msvideo": ".avi",
@@ -88,6 +89,8 @@ const MIME_BY_EXT: Record<string, string> = {
 
 const AMBIGUOUS_VIDEO_MIME_BY_AUDIO_MIME: Readonly<Record<string, string>> = {
   "audio/mp4": "video/mp4",
+  "audio/x-m4a": "video/mp4",
+  "audio/m4a": "video/mp4",
   "audio/webm": "video/webm",
 };
 
@@ -243,9 +246,9 @@ export async function detectMime(opts: {
     : (sniffed ?? extMime);
   // file-type defaults these containers to video without parsing their tracks.
   // Preserve a concrete audio hint only for those documented ambiguous results.
-  const audioContainerHint = mimeHints.find(
-    (mime) => AMBIGUOUS_VIDEO_MIME_BY_AUDIO_MIME[mime] === inferred,
-  );
+  const audioContainerHint =
+    mimeHints.find((mime) => AMBIGUOUS_VIDEO_MIME_BY_AUDIO_MIME[mime] === inferred) ??
+    (extMime && AMBIGUOUS_VIDEO_MIME_BY_AUDIO_MIME[extMime] === inferred ? extMime : undefined);
   if (audioContainerHint) {
     return audioContainerHint;
   }
