Add muxer/demuxer support for E-AC-3 and Dolby Atmos (E-AC-3 JOC)

libraries/extractor/src/main/java/androidx/media3/extractor/mp4/BoxParser.java

@@ -2251,9 +2251,18 @@ private static void parseAudioSampleEntry(
         out.format =
             Ac3Util.parseAc3AnnexFFormat(parent, Integer.toString(trackId), language, drmInitData);
       } else if (childAtomType == Mp4Box.TYPE_dec3) {
+        // Preserve the raw EC3SpecificBox body as csd-0 so that downstream components (e.g. the
+        // muxer for stream-copy) can rewrite the dec3 box.
+        int childAtomBodySize = childAtomSize - Mp4Box.HEADER_SIZE;
+        byte[] initializationDataBytes = new byte[childAtomBodySize];
+        parent.setPosition(Mp4Box.HEADER_SIZE + childPosition);
+        parent.readBytes(initializationDataBytes, /* offset= */ 0, childAtomBodySize);
         parent.setPosition(Mp4Box.HEADER_SIZE + childPosition);
         out.format =
-            Ac3Util.parseEAc3AnnexFFormat(parent, Integer.toString(trackId), language, drmInitData);
+            Ac3Util.parseEAc3AnnexFFormat(parent, Integer.toString(trackId), language, drmInitData)
+                .buildUpon()
+                .setInitializationData(ImmutableList.of(initializationDataBytes))
+                .build();
       } else if (childAtomType == Mp4Box.TYPE_dac4) {
         parent.setPosition(Mp4Box.HEADER_SIZE + childPosition);
         out.format =

libraries/muxer/src/main/java/androidx/media3/muxer/Boxes.java

@@ -781,6 +781,9 @@ public static ByteBuffer codecSpecificBox(Format format) {
         return dOpsBox(format);
       case MimeTypes.AUDIO_IAMF:
         return iacbBox(format);
+      case MimeTypes.AUDIO_E_AC3:
+      case MimeTypes.AUDIO_E_AC3_JOC:
+        return dec3Box(format);
       case MimeTypes.AUDIO_RAW:
         return ByteBuffer.allocate(0); // No codec specific box for raw audio.
       case MimeTypes.VIDEO_H263:
@@ -1643,6 +1646,21 @@ private static ByteBuffer av1CBox(Format format) {
     return BoxUtils.wrapIntoBox("av1C", ByteBuffer.wrap(csd0));
   }
 
+  /**
+   * Returns the dec3 box (EC3SpecificBox) for E-AC-3 / E-AC-3 JOC (Dolby Atmos) audio.
+   *
+   * <p>The dec3 box body is the codec-specific configuration, carried verbatim in csd-0. Stream-copy
+   * callers must populate {@code format.initializationData} with the source track's raw dec3 payload;
+   * the MP4 extractor populates this from the source {@code dec3} box.
+   */
+  private static ByteBuffer dec3Box(Format format) {
+    checkArgument(
+        !format.initializationData.isEmpty(), "csd-0 (dec3 payload) not found for dec3 box.");
+    byte[] csd0 = format.initializationData.get(0);
+    checkArgument(csd0.length > 0, "csd-0 is empty for dec3 box.");
+    return BoxUtils.wrapIntoBox("dec3", ByteBuffer.wrap(csd0));
+  }
+
   /** Returns a dvcC/dvwC/dvvC vision box which will be included in dolby vision box. */
   private static ByteBuffer doviBox(int profile, byte[] csd) {
     checkArgument(csd.length > 0, "csd is empty for dovi box.");
@@ -1847,6 +1865,9 @@ private static String codecSpecificFourcc(Format format) {
         return "Opus";
       case MimeTypes.AUDIO_IAMF:
         return "iamf";
+      case MimeTypes.AUDIO_E_AC3:
+      case MimeTypes.AUDIO_E_AC3_JOC:
+        return "ec-3";
       case MimeTypes.AUDIO_RAW:
         if (format.pcmEncoding == C.ENCODING_PCM_16BIT) {
           return "sowt";

libraries/muxer/src/main/java/androidx/media3/muxer/FragmentedMp4Muxer.java

@@ -62,6 +62,8 @@
  *         <li>Opus
  *         <li>Vorbis
  *         <li>Raw Audio
+ *         <li>E-AC-3 (Dolby Digital Plus)
+ *         <li>E-AC-3 JOC (Dolby Atmos)
  *       </ul>
  *   <li>Metadata
  * </ul>
@@ -180,7 +182,9 @@ public FragmentedMp4Muxer build() {
           MimeTypes.AUDIO_IAMF,
           MimeTypes.AUDIO_OPUS,
           MimeTypes.AUDIO_VORBIS,
-          MimeTypes.AUDIO_RAW);
+          MimeTypes.AUDIO_RAW,
+          MimeTypes.AUDIO_E_AC3,
+          MimeTypes.AUDIO_E_AC3_JOC);
 
   // LINT.ThenChange(Boxes.java:codec_specific_boxes)
 

libraries/muxer/src/main/java/androidx/media3/muxer/Mp4Muxer.java

@@ -84,6 +84,8 @@
  *         <li>Opus
  *         <li>Vorbis
  *         <li>Raw Audio
+ *         <li>E-AC-3 (Dolby Digital Plus)
+ *         <li>E-AC-3 JOC (Dolby Atmos)
  *       </ul>
  *   <li>Metadata
  * </ul>
@@ -414,6 +416,8 @@ public Mp4Muxer build() {
           MimeTypes.AUDIO_OPUS,
           MimeTypes.AUDIO_VORBIS,
           MimeTypes.AUDIO_RAW,
+          MimeTypes.AUDIO_E_AC3,
+          MimeTypes.AUDIO_E_AC3_JOC,
           MimeTypes.AUDIO_IAMF);
 
   // LINT.ThenChange(Boxes.java:codec_specific_boxes)

libraries/muxer/src/test/java/androidx/media3/muxer/BoxesTest.java

@@ -403,6 +403,93 @@ public void createAudioSampleEntryBox_forVorbis_matchesExpected() throws Excepti
         MuxerTestUtil.getExpectedMp4DumpFilePath("audio_sample_entry_box_vorbis"));
   }
 
+  @Test
+  public void createCodecSpecificBox_forEAc3_wrapsDec3PayloadVerbatim() {
+    byte[] payload = new byte[] {0x0F, (byte) 0x80, 0x00};
+    Format format =
+        FAKE_AUDIO_FORMAT
+            .buildUpon()
+            .setSampleMimeType(MimeTypes.AUDIO_E_AC3)
+            .setInitializationData(ImmutableList.of(payload))
+            .build();
+
+    ByteBuffer box = Boxes.codecSpecificBox(format);
+
+    // Box layout: 4 bytes size + 4 bytes "dec3" type + payload bytes.
+    assertThat(box.remaining()).isEqualTo(8 + payload.length);
+    byte[] typeBytes = new byte[4];
+    box.position(4);
+    box.get(typeBytes);
+    assertThat(new String(typeBytes)).isEqualTo("dec3");
+    byte[] actual = new byte[payload.length];
+    box.get(actual);
+    assertThat(actual).isEqualTo(payload);
+  }
+
+  @Test
+  public void createCodecSpecificBox_forEAc3Joc_wrapsDec3PayloadVerbatim() {
+    byte[] payload = new byte[] {0x0F, (byte) 0x80, 0x00};
+    Format format =
+        FAKE_AUDIO_FORMAT
+            .buildUpon()
+            .setSampleMimeType(MimeTypes.AUDIO_E_AC3_JOC)
+            .setInitializationData(ImmutableList.of(payload))
+            .build();
+
+    ByteBuffer box = Boxes.codecSpecificBox(format);
+
+    assertThat(box.remaining()).isEqualTo(8 + payload.length);
+    byte[] typeBytes = new byte[4];
+    box.position(4);
+    box.get(typeBytes);
+    assertThat(new String(typeBytes)).isEqualTo("dec3");
+    byte[] actual = new byte[payload.length];
+    box.get(actual);
+    assertThat(actual).isEqualTo(payload);
+  }
+
+  @Test
+  public void createCodecSpecificBox_forEAc3WithNoCsd_throws() {
+    Format format =
+        FAKE_AUDIO_FORMAT
+            .buildUpon()
+            .setSampleMimeType(MimeTypes.AUDIO_E_AC3)
+            .setInitializationData(ImmutableList.of())
+            .build();
+
+    assertThrows(IllegalArgumentException.class, () -> Boxes.codecSpecificBox(format));
+  }
+
+  @Test
+  public void createCodecSpecificBox_forEAc3WithEmptyCsd_throws() {
+    Format format =
+        FAKE_AUDIO_FORMAT
+            .buildUpon()
+            .setSampleMimeType(MimeTypes.AUDIO_E_AC3)
+            .setInitializationData(ImmutableList.of(new byte[0]))
+            .build();
+
+    assertThrows(IllegalArgumentException.class, () -> Boxes.codecSpecificBox(format));
+  }
+
+  @Test
+  public void createAudioSampleEntryBox_forEAc3_usesEc3Fourcc() {
+    Format format =
+        FAKE_AUDIO_FORMAT
+            .buildUpon()
+            .setSampleMimeType(MimeTypes.AUDIO_E_AC3)
+            .setInitializationData(ImmutableList.of(new byte[] {0x0F, (byte) 0x80, 0x00}))
+            .build();
+
+    ByteBuffer box = Boxes.audioSampleEntry(format);
+
+    // Sample entry box layout: 4 bytes size + 4 bytes fourcc + ...
+    byte[] fourccBytes = new byte[4];
+    box.position(4);
+    box.get(fourccBytes);
+    assertThat(Util.fromUtf8Bytes(fourccBytes)).isEqualTo("ec-3");
+  }
+
   @Test
   public void createAudioSampleEntryBox_withUnknownAudioFormat_throws() {
     // The audio format contains an unknown MIME type.

libraries/muxer/src/test/java/androidx/media3/muxer/Mp4MuxerEndToEndParameterizedTest.java

@@ -66,6 +66,8 @@ public class Mp4MuxerEndToEndParameterizedTest {
   private static final String OPUS_OGG = "ogg/bbb_6ch_8kHz_opus.ogg";
   private static final String VORBIS_OGG = "ogg/bbb_1ch_16kHz_q10_vorbis.ogg";
   private static final String RAW_WAV = "wav/bbb_2ch_44kHz.wav";
+  private static final String EAC3_MP4 = "mp4/sample_eac3.mp4";
+  private static final String EAC3_JOC_MP4 = "mp4/sample_eac3joc.mp4";
 
   public static final String MEDIA_ASSET_DIRECTORY = "asset:///media/";
 
@@ -91,7 +93,9 @@ public static ImmutableList<String> mediaSamples() {
         IAMF_OPUS_MP4,
         OPUS_OGG,
         VORBIS_OGG,
-        RAW_WAV);
+        RAW_WAV,
+        EAC3_MP4,
+        EAC3_JOC_MP4);
   }
 
   @Parameter public @MonotonicNonNull String inputFile;

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3.mp4.0.dump

@@ -21,6 +21,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086526, modification time=3662086526, timescale=48000]
+    initializationData:
+      data = length 5, hash 38742EF
   sample 0:
     time = 0
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3.mp4.1.dump

@@ -21,6 +21,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086526, modification time=3662086526, timescale=48000]
+    initializationData:
+      data = length 5, hash 38742EF
   sample 0:
     time = 576000
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3.mp4.2.dump

@@ -21,6 +21,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086526, modification time=3662086526, timescale=48000]
+    initializationData:
+      data = length 5, hash 38742EF
   sample 0:
     time = 1152000
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3.mp4.3.dump

@@ -21,6 +21,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086526, modification time=3662086526, timescale=48000]
+    initializationData:
+      data = length 5, hash 38742EF
   sample 0:
     time = 1696000
     flags = 536870913

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3.mp4.unknown_length.dump

@@ -21,6 +21,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086526, modification time=3662086526, timescale=48000]
+    initializationData:
+      data = length 5, hash 38742EF
   sample 0:
     time = 0
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3_fragmented.mp4.0.dump

@@ -19,6 +19,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086719, modification time=3662086719, timescale=1000]
+    initializationData:
+      data = length 5, hash 38742EF
   sample 0:
     time = 0
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3_fragmented.mp4.1.dump

@@ -19,6 +19,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086719, modification time=3662086719, timescale=1000]
+    initializationData:
+      data = length 5, hash 38742EF
   sample 0:
     time = 576000
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3_fragmented.mp4.2.dump

@@ -19,6 +19,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086719, modification time=3662086719, timescale=1000]
+    initializationData:
+      data = length 5, hash 38742EF
   sample 0:
     time = 1152000
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3_fragmented.mp4.3.dump

@@ -19,6 +19,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086719, modification time=3662086719, timescale=1000]
+    initializationData:
+      data = length 5, hash 38742EF
   sample 0:
     time = 1696000
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3_fragmented.mp4.unknown_length.dump

@@ -19,6 +19,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086719, modification time=3662086719, timescale=1000]
+    initializationData:
+      data = length 5, hash 38742EF
   sample 0:
     time = 0
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3joc.mp4.0.dump

@@ -21,6 +21,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086940, modification time=3662086940, timescale=48000]
+    initializationData:
+      data = length 7, hash 8BA78FD3
   sample 0:
     time = 0
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3joc.mp4.1.dump

@@ -21,6 +21,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086940, modification time=3662086940, timescale=48000]
+    initializationData:
+      data = length 7, hash 8BA78FD3
   sample 0:
     time = 672000
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3joc.mp4.2.dump

@@ -21,6 +21,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086940, modification time=3662086940, timescale=48000]
+    initializationData:
+      data = length 7, hash 8BA78FD3
   sample 0:
     time = 1344000
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3joc.mp4.3.dump

@@ -21,6 +21,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086940, modification time=3662086940, timescale=48000]
+    initializationData:
+      data = length 7, hash 8BA78FD3
   sample 0:
     time = 2016000
     flags = 536870913

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3joc.mp4.unknown_length.dump

@@ -21,6 +21,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086940, modification time=3662086940, timescale=48000]
+    initializationData:
+      data = length 7, hash 8BA78FD3
   sample 0:
     time = 0
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3joc_fragmented.mp4.0.dump

@@ -19,6 +19,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086977, modification time=3662086977, timescale=1000]
+    initializationData:
+      data = length 7, hash 8BA78FD3
   sample 0:
     time = 0
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3joc_fragmented.mp4.1.dump

@@ -19,6 +19,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086977, modification time=3662086977, timescale=1000]
+    initializationData:
+      data = length 7, hash 8BA78FD3
   sample 0:
     time = 672000
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3joc_fragmented.mp4.2.dump

@@ -19,6 +19,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086977, modification time=3662086977, timescale=1000]
+    initializationData:
+      data = length 7, hash 8BA78FD3
   sample 0:
     time = 1344000
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3joc_fragmented.mp4.3.dump

@@ -19,6 +19,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086977, modification time=3662086977, timescale=1000]
+    initializationData:
+      data = length 7, hash 8BA78FD3
   sample 0:
     time = 2016000
     flags = 1

libraries/test_data/src/test/assets/extractordumps/mp4/sample_eac3joc_fragmented.mp4.unknown_length.dump

@@ -19,6 +19,8 @@ track 0:
     sampleRate = 48000
     language = und
     metadata = entries=[Mp4Timestamp: creation time=3662086977, modification time=3662086977, timescale=1000]
+    initializationData:
+      data = length 7, hash 8BA78FD3
   sample 0:
     time = 0
     flags = 1