diff --git a/lib/util/mp4_box_parsers.js b/lib/util/mp4_box_parsers.js
index 6abdcbc1a4..23a4e7e656 100644
--- a/lib/util/mp4_box_parsers.js
+++ b/lib/util/mp4_box_parsers.js
@@ -301,11 +301,45 @@ shaka.util.Mp4BoxParsers = class {
    */
   static parseESDS(reader) {
     let codec = 'mp4a';
-    reader.skip(11);
-    codec += '.' + this.toHex_(reader.readUint8());
-    // this value is only a single digit
-    codec += '.' +
-        this.toHex_((reader.readUint8() >>> 2) & 0x3f).replace(/^0/, '');
+    let tag;
+    let oti;
+    while (reader.hasMoreData()) {
+      tag = reader.readUint8();
+      let byteRead = reader.readUint8();
+      while (byteRead & 0x80) {
+        byteRead = reader.readUint8();
+      }
+      if (tag == 0x03) {
+        reader.readUint16();
+        const flags = reader.readUint8();
+        if (flags & 0x80) {
+          reader.readUint16();
+        }
+        if (flags & 0x40) {
+          reader.skip(reader.readUint8());
+        }
+        if (flags & 0x20) {
+          reader.readUint16();
+        }
+      } else if (tag == 0x04) {
+        oti = reader.readUint8();
+        reader.skip(12);
+      } else if (tag == 0x05) {
+        break;
+      }
+    }
+    if (oti) {
+      codec += '.' + this.toHex_(oti);
+      if (tag == 0x05 && reader.hasMoreData()) {
+        const firstData = reader.readUint8();
+        let audioObjectType = (firstData & 0xF8) >> 3;
+        if (audioObjectType === 31 && reader.hasMoreData()) {
+          audioObjectType = 32 + ((firstData & 0x7) << 3) +
+              ((reader.readUint8() & 0xE0) >> 5);
+        }
+        codec += '.' + audioObjectType;
+      }
+    }
     return {codec};
   }
 
diff --git a/test/test/assets/audio-xhe-aac.mp4 b/test/test/assets/audio-xhe-aac.mp4
new file mode 100644
index 0000000000..47d432661e
Binary files /dev/null and b/test/test/assets/audio-xhe-aac.mp4 differ
diff --git a/test/util/mp4_box_parsers_unit.js b/test/util/mp4_box_parsers_unit.js
index fab41ee073..41a8827af8 100644
--- a/test/util/mp4_box_parsers_unit.js
+++ b/test/util/mp4_box_parsers_unit.js
@@ -8,18 +8,24 @@ describe('Mp4BoxParsers', () => {
   const videoInitSegmentUri = '/base/test/test/assets/sintel-video-init.mp4';
   const videoSegmentUri = '/base/test/test/assets/sintel-video-segment.mp4';
 
+  const audioInitSegmentXheAacUri = '/base/test/test/assets/audio-xhe-aac.mp4';
+
   /** @type {!ArrayBuffer} */
   let videoInitSegment;
   /** @type {!ArrayBuffer} */
   let videoSegment;
+  /** @type {!ArrayBuffer} */
+  let audioInitSegmentXheAac;
 
   beforeAll(async () => {
     const responses = await Promise.all([
       shaka.test.Util.fetch(videoInitSegmentUri),
       shaka.test.Util.fetch(videoSegmentUri),
+      shaka.test.Util.fetch(audioInitSegmentXheAacUri),
     ]);
     videoInitSegment = responses[0];
     videoSegment = responses[1];
+    audioInitSegmentXheAac = responses[2];
   });
 
   it('parses init segment', () => {
@@ -157,6 +163,36 @@ describe('Mp4BoxParsers', () => {
     expect(baseMediaDecodeTime).toBe(expectedBaseMediaDecodeTime);
   });
 
+  it('parses ESDS box for xHE-AAC segment', () => {
+    let channelCount;
+    let sampleRate;
+    let codec;
+
+    const Mp4Parser = shaka.util.Mp4Parser;
+    new Mp4Parser()
+        .box('moov', Mp4Parser.children)
+        .box('trak', Mp4Parser.children)
+        .box('mdia', Mp4Parser.children)
+        .box('minf', Mp4Parser.children)
+        .box('stbl', Mp4Parser.children)
+        .fullBox('stsd', Mp4Parser.sampleDescription)
+        .box('mp4a', (box) => {
+          const parsedMP4ABox = shaka.util.Mp4BoxParsers.parseMP4A(box.reader);
+          channelCount = parsedMP4ABox.channelCount;
+          sampleRate = parsedMP4ABox.sampleRate;
+          if (box.reader.hasMoreData()) {
+            Mp4Parser.children(box);
+          }
+        })
+        .box('esds', (box) => {
+          const parsedESDSBox = shaka.util.Mp4BoxParsers.parseESDS(box.reader);
+          codec = parsedESDSBox.codec;
+        }).parse(audioInitSegmentXheAac, /* partialOkay= */ false);
+    expect(channelCount).toBe(2);
+    expect(sampleRate).toBe(48000);
+    expect(codec).toBe('mp4a.40.42');
+  });
+
   /**
    *
    * Explanation on the Uint8Array: