Fixed issue with ALAC 20-bit files not having PCM format detected

libraries/common/src/main/java/androidx/media3/common/C.java

@@ -179,7 +179,8 @@ private C() {}
   /**
    * Represents an audio encoding, or an invalid or unset value. One of {@link Format#NO_VALUE},
    * {@link #ENCODING_INVALID}, {@link #ENCODING_PCM_8BIT}, {@link #ENCODING_PCM_16BIT}, {@link
-   * #ENCODING_PCM_16BIT_BIG_ENDIAN}, {@link #ENCODING_PCM_24BIT}, {@link
+   * #ENCODING_PCM_16BIT_BIG_ENDIAN}, {@link #ENCODING_PCM_20BIT}, {@link
+   * #ENCODING_PCM_20BIT_BIG_ENDIAN}, {@link #ENCODING_PCM_24BIT}, {@link
    * #ENCODING_PCM_24BIT_BIG_ENDIAN}, {@link #ENCODING_PCM_32BIT}, {@link
    * #ENCODING_PCM_32BIT_BIG_ENDIAN}, {@link #ENCODING_PCM_FLOAT}, {@link #ENCODING_PCM_DOUBLE},
    * {@link #ENCODING_MP3}, {@link #ENCODING_AC3}, {@link #ENCODING_E_AC3}, {@link
@@ -195,6 +196,8 @@ private C() {}
     ENCODING_PCM_8BIT,
     ENCODING_PCM_16BIT,
     ENCODING_PCM_16BIT_BIG_ENDIAN,
+    ENCODING_PCM_20BIT,
+    ENCODING_PCM_20BIT_BIG_ENDIAN,
     ENCODING_PCM_24BIT,
     ENCODING_PCM_24BIT_BIG_ENDIAN,
     ENCODING_PCM_32BIT,
@@ -224,7 +227,8 @@ private C() {}
   /**
    * Represents a PCM audio encoding, or an invalid or unset value. One of {@link Format#NO_VALUE},
    * {@link #ENCODING_INVALID}, {@link #ENCODING_PCM_8BIT}, {@link #ENCODING_PCM_16BIT}, {@link
-   * #ENCODING_PCM_16BIT_BIG_ENDIAN}, {@link #ENCODING_PCM_24BIT}, {@link
+   * #ENCODING_PCM_16BIT_BIG_ENDIAN}, {@link #ENCODING_PCM_20BIT}, {@link
+   * #ENCODING_PCM_20BIT_BIG_ENDIAN}, {@link #ENCODING_PCM_24BIT}, {@link
    * #ENCODING_PCM_24BIT_BIG_ENDIAN}, {@link #ENCODING_PCM_32BIT}, {@link
    * #ENCODING_PCM_32BIT_BIG_ENDIAN}, {@link #ENCODING_PCM_FLOAT}, {@link #ENCODING_PCM_DOUBLE}.
    */
@@ -237,6 +241,8 @@ private C() {}
     ENCODING_PCM_8BIT,
     ENCODING_PCM_16BIT,
     ENCODING_PCM_16BIT_BIG_ENDIAN,
+    ENCODING_PCM_20BIT,
+    ENCODING_PCM_20BIT_BIG_ENDIAN,
     ENCODING_PCM_24BIT,
     ENCODING_PCM_24BIT_BIG_ENDIAN,
     ENCODING_PCM_32BIT,
@@ -258,6 +264,26 @@ private C() {}
   /** Like {@link #ENCODING_PCM_16BIT}, but with the bytes in big endian order. */
   @UnstableApi public static final int ENCODING_PCM_16BIT_BIG_ENDIAN = 0x10000000;
 
+  /**
+   * PCM encoding with 20 bits per sample.
+   *
+   * <p>Note that this is not generally supported or used and just exists to signal that a
+   * compressed audio track contains 20-bit PCM resolution. A decoder for said track should convert
+   * the audio to the closest higher format (such as {@link #ENCODING_PCM_24BIT}) instead of
+   * attempting to output 20-bit PCM.
+   */
+  @UnstableApi public static final int ENCODING_PCM_20BIT = 0x80000000;
+
+  /**
+   * Like {@link #ENCODING_PCM_20BIT} but with the bytes in big endian order.
+   *
+   * <p>Note that this is not generally supported or used and just exists to signal that a
+   * compressed audio track contains 20-bit PCM resolution. A decoder for said track should convert
+   * the audio to the closest higher format (such as {@link #ENCODING_PCM_24BIT}) instead of
+   * attempting to output 20-bit PCM.
+   */
+  @UnstableApi public static final int ENCODING_PCM_20BIT_BIG_ENDIAN = 0x90000000;
+
   /** PCM encoding with 24 bits per sample. */
   public static final int ENCODING_PCM_24BIT = AudioFormat.ENCODING_PCM_24BIT_PACKED;
 

libraries/common/src/main/java/androidx/media3/common/util/MediaFormatUtil.java

@@ -504,6 +504,8 @@ private static void maybeSetPcmEncoding(
         break;
       case Format.NO_VALUE:
       case C.ENCODING_PCM_16BIT_BIG_ENDIAN:
+      case C.ENCODING_PCM_20BIT:
+      case C.ENCODING_PCM_20BIT_BIG_ENDIAN:
       case C.ENCODING_PCM_24BIT_BIG_ENDIAN:
       case C.ENCODING_PCM_32BIT_BIG_ENDIAN:
       case C.ENCODING_PCM_DOUBLE:

libraries/common/src/main/java/androidx/media3/common/util/Util.java

@@ -2380,6 +2380,10 @@ public static Format getPcmFormat(AudioProcessor.AudioFormat audioFormat) {
   /**
    * Converts a sample bit depth to a corresponding little-endian integer PCM encoding constant.
    *
+   * <p>Note that this method does not support encodings which are not a multiple of bytes, such as
+   * 20-bit PCM, as those cannot be handled in the same way as linear PCM encodings which are
+   * multiples of bytes. It's recommended to convert such a PCM format to the closest higher format.
+   *
    * @param bitDepth The bit depth. Supported values are 8, 16, 24 and 32.
    * @return The corresponding encoding. One of {@link C#ENCODING_PCM_8BIT}, {@link
    *     C#ENCODING_PCM_16BIT}, {@link C#ENCODING_PCM_24BIT} and {@link C#ENCODING_PCM_32BIT}. If
@@ -2393,6 +2397,10 @@ public static Format getPcmFormat(AudioProcessor.AudioFormat audioFormat) {
   /**
    * Converts a sample bit depth and byte order to a corresponding integer PCM encoding constant.
    *
+   * <p>Note that this method does not support encodings which are not a multiple of bytes, such as
+   * 20-bit PCM, as those cannot be handled in the same way as linear PCM encodings which are
+   * multiples of bytes. It's recommended to convert such a PCM format to the closest higher format.
+   *
    * @param bitDepth The bit depth. Supported values are 8, 16, 24 and 32.
    * @param byteOrder The byte order.
    * @return The corresponding integer PCM encoding. If the bit depth is unsupported then {@link
@@ -2424,6 +2432,10 @@ public static Format getPcmFormat(AudioProcessor.AudioFormat audioFormat) {
   /**
    * Returns whether {@code encoding} is one of the linear PCM encodings.
    *
+   * <p>Note that this will return false for encodings which are not a multiple of bytes, such as
+   * 20-bit PCM, as those cannot be handled in the same way as linear PCM encodings which are
+   * multiples of bytes. It's recommended to convert such a PCM format to the closest higher format.
+   *
    * @param encoding The encoding of the audio data.
    * @return Whether the encoding is one of the PCM encodings.
    */
@@ -2443,6 +2455,10 @@ public static boolean isEncodingLinearPcm(@C.Encoding int encoding) {
   /**
    * Returns whether {@code encoding} is high resolution (&gt; 16-bit) PCM.
    *
+   * <p>Note that this will return false for encodings which are not a multiple of bytes, such as
+   * 20-bit PCM, as those cannot be handled in the same way as linear PCM encodings which are
+   * multiples of bytes. It's recommended to convert such a PCM format to the closest higher format.
+   *
    * @param encoding The encoding of the audio data.
    * @return Whether the encoding is high resolution PCM.
    */
@@ -2616,35 +2632,60 @@ public static int getPcmFrameSize(@C.PcmEncoding int pcmEncoding, int channelCou
   }
 
   /**
-   * Returns the byte depth for audio with the specified encoding.
+   * Returns the bit depth for audio with the specified encoding.
    *
    * @param pcmEncoding The encoding of the audio data.
-   * @return The byte depth of the audio.
+   * @return The bit depth of the audio.
    */
   @UnstableApi
-  public static int getByteDepth(@C.PcmEncoding int pcmEncoding) {
+  public static int getBitDepth(@C.PcmEncoding int pcmEncoding) {
     switch (pcmEncoding) {
       case C.ENCODING_PCM_8BIT:
-        return 1;
+        return 8;
       case C.ENCODING_PCM_16BIT:
       case C.ENCODING_PCM_16BIT_BIG_ENDIAN:
-        return 2;
+        return 16;
+      case C.ENCODING_PCM_20BIT:
+      case C.ENCODING_PCM_20BIT_BIG_ENDIAN:
+        return 20;
       case C.ENCODING_PCM_24BIT:
       case C.ENCODING_PCM_24BIT_BIG_ENDIAN:
-        return 3;
+        return 24;
       case C.ENCODING_PCM_32BIT:
       case C.ENCODING_PCM_32BIT_BIG_ENDIAN:
       case C.ENCODING_PCM_FLOAT:
-        return 4;
+        return 32;
       case C.ENCODING_PCM_DOUBLE:
-        return 8;
+        return 64;
       case C.ENCODING_INVALID:
       case Format.NO_VALUE:
       default:
         throw new IllegalArgumentException();
     }
   }
 
+  /**
+   * Returns the byte depth for audio with the specified encoding.
+   *
+   * <p>This will throw for encodings which are not a multiple of bytes, such as 20-bit PCM.
+   *
+   * @param pcmEncoding The encoding of the audio data.
+   * @return The byte depth of the audio.
+   * @see #getBitDepth(int)
+   */
+  @UnstableApi
+  public static int getByteDepth(@C.PcmEncoding int pcmEncoding) {
+    int bitDepth = getBitDepth(pcmEncoding);
+    if ((bitDepth % C.BITS_PER_BYTE) != 0) {
+      throw new IllegalArgumentException(
+          "Bit depth "
+              + bitDepth
+              + " cannot be represented as byte"
+              + " depth. Use getBitDepth() instead.");
+    }
+    return bitDepth / C.BITS_PER_BYTE;
+  }
+
   /** Returns the {@link C.AudioUsage} corresponding to the specified {@link C.StreamType}. */
   @UnstableApi
   public static @C.AudioUsage int getAudioUsageForStreamType(@C.StreamType int streamType) {

libraries/exoplayer/src/main/java/androidx/media3/exoplayer/audio/DefaultAudioSink.java

@@ -1828,6 +1828,8 @@ private FormatConfig getFormatConfig(Format format, int preferredBufferSize) {
         return OpusUtil.parseOggPacketAudioSampleCount(buffer);
       case C.ENCODING_PCM_16BIT:
       case C.ENCODING_PCM_16BIT_BIG_ENDIAN:
+      case C.ENCODING_PCM_20BIT:
+      case C.ENCODING_PCM_20BIT_BIG_ENDIAN:
       case C.ENCODING_PCM_24BIT:
       case C.ENCODING_PCM_24BIT_BIG_ENDIAN:
       case C.ENCODING_PCM_32BIT:

libraries/exoplayer/src/main/java/androidx/media3/exoplayer/util/EventLogger.java

@@ -821,6 +821,10 @@ private static String encodingAsString(@C.Encoding int encoding) {
         return "pcm-16";
       case C.ENCODING_PCM_16BIT_BIG_ENDIAN:
         return "pcm-16be";
+      case C.ENCODING_PCM_20BIT:
+        return "pcm-20";
+      case C.ENCODING_PCM_20BIT_BIG_ENDIAN:
+        return "pcm-20be";
       case C.ENCODING_PCM_24BIT:
         return "pcm-24";
       case C.ENCODING_PCM_24BIT_BIG_ENDIAN:

libraries/extractor/src/main/java/androidx/media3/extractor/ExtractorUtil.java

@@ -163,6 +163,8 @@ public static int getMaximumEncodedRateBytesPerSecond(@C.Encoding int encoding)
         return OpusUtil.MAX_BYTES_PER_SECOND;
       case C.ENCODING_PCM_16BIT:
       case C.ENCODING_PCM_16BIT_BIG_ENDIAN:
+      case C.ENCODING_PCM_20BIT:
+      case C.ENCODING_PCM_20BIT_BIG_ENDIAN:
       case C.ENCODING_PCM_24BIT:
       case C.ENCODING_PCM_24BIT_BIG_ENDIAN:
       case C.ENCODING_PCM_32BIT:

libraries/extractor/src/main/java/androidx/media3/extractor/WavUtil.java

@@ -72,13 +72,15 @@ public static int getTypeForPcmEncoding(@C.PcmEncoding int pcmEncoding) {
     switch (pcmEncoding) {
       case C.ENCODING_PCM_8BIT:
       case C.ENCODING_PCM_16BIT:
+      case C.ENCODING_PCM_20BIT:
       case C.ENCODING_PCM_24BIT:
       case C.ENCODING_PCM_32BIT:
         return TYPE_PCM;
       case C.ENCODING_PCM_FLOAT:
         return TYPE_FLOAT;
       // TYPE_PCM is little endian so big endian formats don't match.
       case C.ENCODING_PCM_16BIT_BIG_ENDIAN:
+      case C.ENCODING_PCM_20BIT_BIG_ENDIAN:
       case C.ENCODING_PCM_24BIT_BIG_ENDIAN:
       case C.ENCODING_PCM_32BIT_BIG_ENDIAN:
       case C.ENCODING_INVALID:

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

@@ -2376,7 +2376,13 @@ private static void parseAudioSampleEntry(
         sampleRate = parsedAlacConfig[0];
         channelCount = parsedAlacConfig[1];
         int bitDepth = parsedAlacConfig[2];
-        pcmEncoding = Util.getPcmEncoding(bitDepth);
+        // getPcmEncoding() does not support encodings which are not a multiple of bytes, such as
+        // 20-bit PCM, as those cannot be handled in the same way as linear PCM encodings which are
+        // multiples of bytes. These formats are also not supported by any part of media3's PCM
+        // handling pipeline. The reason this constant exists is to be able to signal that the
+        // compressed audio (in this case, ALAC) has 20-bit PCM precision, but a decoder is expected
+        // to output a supported format such as 24-bit PCM instead.
+        pcmEncoding = bitDepth == 20 ? C.ENCODING_PCM_20BIT : Util.getPcmEncoding(bitDepth);
         initializationData = ImmutableList.of(initializationDataBytes);
       } else if (childAtomType == Mp4Box.TYPE_iacb) {
         parent.setPosition(

libraries/extractor/src/test/java/androidx/media3/extractor/mp4/Mp4ExtractorParameterizedTest.java

@@ -308,6 +308,11 @@ public void mp4SampleWithAlac() throws Exception {
     assertExtractorBehavior("media/mp4/sample_alac.mp4", /* peekLimit= */ 50);
   }
 
+  @Test
+  public void mp4SampleWithAlac20Bit() throws Exception {
+    assertExtractorBehavior("media/mp4/sample_alac_20bit.mp4", /* peekLimit= */ 4096);
+  }
+
   @Test
   public void mp4SampleWithFixedRechunkAndNoElst() throws Exception {
     assertExtractorBehavior("media/mp4/sample_fixed_rechunk_no_elst.mp4", /* peekLimit= */ 44);

libraries/test_data/src/test/assets/extractordumps/mp4/sample_alac_20bit.mp4.0.dump

@@ -0,0 +1,74 @@
+seekMap:
+  isSeekable = true
+  duration = 1021678
+  getPosition(0) = [[timeUs=0, position=4096]]
+  getPosition(1) = [[timeUs=1, position=4096]]
+  getPosition(510839) = [[timeUs=510839, position=21666]]
+  getPosition(1021678) = [[timeUs=1021678, position=44223]]
+numberOfTracks = 1
+track 0:
+  total output bytes = 40135
+  sample count = 12
+  track duration = 1021678
+  format 0:
+    averageBitrate = 288079
+    id = 1
+    containerMimeType = audio/mp4
+    sampleMimeType = audio/alac
+    maxInputSize = 4105
+    channelCount = 1
+    sampleRate = 44100
+    pcmEncoding = -2147483648
+    encoderPadding = 4096
+    metadata = entries=[Mp4Timestamp: creation time=3853596544, modification time=3853596544, timescale=44100]
+    initializationData:
+      data = length 24, hash 478A29BA
+  sample 0:
+    time = 0
+    flags = 1
+    data = length 1952, hash 298B8810
+  sample 1:
+    time = 92879
+    flags = 1
+    data = length 3898, hash 7734D643
+  sample 2:
+    time = 185759
+    flags = 1
+    data = length 3923, hash 8AAC13A1
+  sample 3:
+    time = 278639
+    flags = 1
+    data = length 3917, hash CFEBDF3E
+  sample 4:
+    time = 371519
+    flags = 1
+    data = length 3880, hash CCDB7EB5
+  sample 5:
+    time = 464399
+    flags = 1
+    data = length 3830, hash 7FB90D6E
+  sample 6:
+    time = 557278
+    flags = 1
+    data = length 3845, hash 4481CE0F
+  sample 7:
+    time = 650158
+    flags = 1
+    data = length 3850, hash 29F6A218
+  sample 8:
+    time = 743038
+    flags = 1
+    data = length 3906, hash A1084987
+  sample 9:
+    time = 835918
+    flags = 1
+    data = length 4075, hash 2D7C7AA4
+  sample 10:
+    time = 928798
+    flags = 1
+    data = length 3051, hash 5D47B269
+  sample 11:
+    time = 1021678
+    flags = 536870913
+    data = length 8, hash B2FBAA0E
+tracksEnded = true

libraries/test_data/src/test/assets/extractordumps/mp4/sample_alac_20bit.mp4.1.dump

@@ -0,0 +1,62 @@
+seekMap:
+  isSeekable = true
+  duration = 1021678
+  getPosition(0) = [[timeUs=0, position=4096]]
+  getPosition(1) = [[timeUs=1, position=4096]]
+  getPosition(510839) = [[timeUs=510839, position=21666]]
+  getPosition(1021678) = [[timeUs=1021678, position=44223]]
+numberOfTracks = 1
+track 0:
+  total output bytes = 30362
+  sample count = 9
+  track duration = 1021678
+  format 0:
+    averageBitrate = 288079
+    id = 1
+    containerMimeType = audio/mp4
+    sampleMimeType = audio/alac
+    maxInputSize = 4105
+    channelCount = 1
+    sampleRate = 44100
+    pcmEncoding = -2147483648
+    encoderPadding = 4096
+    metadata = entries=[Mp4Timestamp: creation time=3853596544, modification time=3853596544, timescale=44100]
+    initializationData:
+      data = length 24, hash 478A29BA
+  sample 0:
+    time = 278639
+    flags = 1
+    data = length 3917, hash CFEBDF3E
+  sample 1:
+    time = 371519
+    flags = 1
+    data = length 3880, hash CCDB7EB5
+  sample 2:
+    time = 464399
+    flags = 1
+    data = length 3830, hash 7FB90D6E
+  sample 3:
+    time = 557278
+    flags = 1
+    data = length 3845, hash 4481CE0F
+  sample 4:
+    time = 650158
+    flags = 1
+    data = length 3850, hash 29F6A218
+  sample 5:
+    time = 743038
+    flags = 1
+    data = length 3906, hash A1084987
+  sample 6:
+    time = 835918
+    flags = 1
+    data = length 4075, hash 2D7C7AA4
+  sample 7:
+    time = 928798
+    flags = 1
+    data = length 3051, hash 5D47B269
+  sample 8:
+    time = 1021678
+    flags = 536870913
+    data = length 8, hash B2FBAA0E
+tracksEnded = true