Support Dolby audio Enhanced AC3 in ISO BMFF (Part 2)

- Box definitions for box type DEC3.
- Parser/muxer changes to support Enhanced-AC3 audio codecs.
- MPD signaling will come in Part 3.

Issue #64

Change-Id: Ifcd5efa1f61b470ec225127925631e4329853259

packager/media/base/audio_stream_info.cc

@@ -33,6 +33,8 @@ std::string AudioCodecToString(AudioCodec audio_codec) {
       return "DTS-";
     case kCodecDTSP:
       return "DTS+";
+    case kCodecEAC3:
+      return "EAC3";
     case kCodecOpus:
       return "Opus";
     case kCodecVorbis:
@@ -115,6 +117,8 @@ std::string AudioStreamInfo::GetCodecString(AudioCodec codec,
       return "dts-";
     case kCodecAC3:
       return "ac-3";
+    case kCodecEAC3:
+      return "ec-3";
     default:
       NOTIMPLEMENTED() << "Codec: " << codec;
       return "unknown";

packager/media/base/audio_stream_info.h

@@ -24,6 +24,7 @@ enum AudioCodec {
   kCodecDTSL,
   kCodecDTSM,
   kCodecDTSP,
+  kCodecEAC3,
   kCodecOpus,
   kCodecVorbis,
 

packager/media/formats/mp4/box_definitions.cc

@@ -8,6 +8,7 @@
 
 #include "packager/base/logging.h"
 #include "packager/media/base/bit_reader.h"
+#include "packager/media/base/macros.h"
 #include "packager/media/formats/mp4/box_buffer.h"
 #include "packager/media/formats/mp4/rcheck.h"
 
@@ -1435,6 +1436,54 @@ uint32_t AC3Specific::ComputeSizeInternal() {
   return HeaderSize() + data.size();
 }
 
+EC3Specific::EC3Specific() : number_independent_substreams(0) {}
+EC3Specific::~EC3Specific() {}
+
+FourCC EC3Specific::BoxType() const { return FOURCC_DEC3; }
+
+bool EC3Specific::ReadWriteInternal(BoxBuffer* buffer) {
+  RCHECK(ReadWriteHeaderInternal(buffer));
+  uint32_t size = buffer->Reading() ? buffer->BytesLeft() : data.size();
+  RCHECK(buffer->ReadWriteVector(&data, size));
+
+  // Skip data rate, read number of independent substreams and parse the
+  // independent substreams.
+  BitReader bit_reader(&data[0], size);
+  RCHECK(bit_reader.SkipBits(13) &&
+         bit_reader.ReadBits(3, &number_independent_substreams));
+
+  // The value of this field is one less than the number of independent
+  // substreams present.
+  ++number_independent_substreams;
+  IndependentSubstream substream;
+  for (size_t i = 0; i < number_independent_substreams; ++i) {
+    RCHECK(bit_reader.ReadBits(2, &substream.sample_rate_code));
+    RCHECK(bit_reader.ReadBits(5, &substream.bit_stream_identification));
+    RCHECK(bit_reader.SkipBits(1));
+    RCHECK(bit_reader.ReadBits(1, &substream.audio_service));
+    RCHECK(bit_reader.ReadBits(3, &substream.bit_stream_mode));
+    RCHECK(bit_reader.ReadBits(3, &substream.audio_coding_mode));
+    RCHECK(bit_reader.ReadBits(1, &substream.lfe_channel_on));
+    RCHECK(bit_reader.SkipBits(3));
+    RCHECK(bit_reader.ReadBits(4, &substream.number_dependent_substreams));
+    if (substream.number_dependent_substreams > 0) {
+      RCHECK(bit_reader.ReadBits(9, &substream.channel_location));
+    } else {
+      RCHECK(bit_reader.SkipBits(1));
+    }
+    independent_substreams.push_back(substream);
+  }
+
+  return true;
+}
+
+uint32_t EC3Specific::ComputeSizeInternal() {
+  // This box is optional. Skip it if not initialized.
+  if (data.empty())
+    return 0;
+  return HeaderSize() + data.size();
+}
+
 AudioSampleEntry::AudioSampleEntry()
     : format(FOURCC_NULL),
       data_reference_index(1),
@@ -1489,6 +1538,7 @@ bool AudioSampleEntry::ReadWriteInternal(BoxBuffer* buffer) {
   RCHECK(buffer->TryReadWriteChild(&esds));
   RCHECK(buffer->TryReadWriteChild(&ddts));
   RCHECK(buffer->TryReadWriteChild(&dac3));
+  RCHECK(buffer->TryReadWriteChild(&dec3));
   return true;
 }
 
@@ -1496,6 +1546,7 @@ uint32_t AudioSampleEntry::ComputeSizeInternal() {
   return HeaderSize() + sizeof(data_reference_index) + sizeof(channelcount) +
          sizeof(samplesize) + sizeof(samplerate) + sinf.ComputeSize() +
          esds.ComputeSize() + ddts.ComputeSize() + dac3.ComputeSize() +
+         dec3.ComputeSize() +
          6 + 8 +  // 6 + 8 bytes reserved.
          4;       // 4 bytes predefined.
 }

packager/media/formats/mp4/box_definitions.h

@@ -315,6 +315,34 @@ struct AC3Specific : Box {
   std::vector<uint8_t> data;
 };
 
+// Independent substream in EC3Specific box.
+struct IndependentSubstream {
+  uint8_t sample_rate_code;             // fscod: 2 bits
+  uint8_t bit_stream_identification;    // bsid: 5 bits
+  // reserved_1: 1 bit
+  uint8_t audio_service;                // asvc: 1 bit
+  uint8_t bit_stream_mode;              // bsmod: 3 bits
+  uint8_t audio_coding_mode;            // acmod: 3 bits
+  uint8_t lfe_channel_on;               // lfeon: 1 bit
+  // reserved_2: 3 bit
+  uint8_t number_dependent_substreams;  // num_dep_sub: 4 bits.
+  // If num_dep_sub > 0, chan_loc is present and the size is 9 bits.
+  // Otherwise, reserved_3 is present and the size is 1 bit.
+  uint16_t channel_location;            // chan_loc: 9 bits.
+  // reserved_3: 1 bit
+};
+
+struct EC3Specific : Box {
+  DECLARE_BOX_METHODS(EC3Specific);
+
+  // Before we know the number of independent substreams, data in EC3Specific
+  // box is store for parsing later.
+  std::vector<uint8_t> data;
+
+  size_t number_independent_substreams;  // num_id_sub: 3 bits.
+  std::vector<IndependentSubstream> independent_substreams;
+};
+
 struct AudioSampleEntry : Box {
   DECLARE_BOX_METHODS(AudioSampleEntry);
   // Returns actual format of this sample entry.
@@ -329,9 +357,11 @@ struct AudioSampleEntry : Box {
   uint32_t samplerate;
 
   ProtectionSchemeInfo sinf;
+
   ElementaryStreamDescriptor esds;
   DTSSpecific ddts;
   AC3Specific dac3;
+  EC3Specific dec3;
 };
 
 struct WebVTTConfigurationBox : Box {

packager/media/formats/mp4/box_definitions_comparison.h

@@ -245,6 +245,11 @@ inline bool operator==(const AC3Specific& lhs,
   return lhs.data == rhs.data;
 }
 
+inline bool operator==(const EC3Specific& lhs,
+                       const EC3Specific& rhs) {
+  return lhs.data == rhs.data;
+}
+
 inline bool operator==(const AudioSampleEntry& lhs,
                        const AudioSampleEntry& rhs) {
   return lhs.format == rhs.format &&
@@ -252,7 +257,7 @@ inline bool operator==(const AudioSampleEntry& lhs,
          lhs.channelcount == rhs.channelcount &&
          lhs.samplesize == rhs.samplesize && lhs.samplerate == rhs.samplerate &&
          lhs.sinf == rhs.sinf && lhs.esds == rhs.esds && lhs.ddts == rhs.ddts &&
-         lhs.dac3 == rhs.dac3;
+         lhs.dac3 == rhs.dac3 && lhs.dec3 == rhs.dec3;
 }
 
 inline bool operator==(const WebVTTConfigurationBox& lhs,

packager/media/formats/mp4/box_definitions_unittest.cc

@@ -390,6 +390,16 @@ class BoxDefinitionsTestGeneral : public testing::Test {
     dac3->data.assign(kAc3Data, kAc3Data + arraysize(kAc3Data));
   }
 
+  void Fill(EC3Specific* dec3) {
+    const uint8_t kEc3Data[] = {0x08, 0x00, 0x20, 0x0f, 0x00};
+    dec3->data.assign(kEc3Data, kEc3Data + arraysize(kEc3Data));
+  }
+
+  void Modify(EC3Specific* dec3) {
+    const uint8_t kEc3Data[] = {0x07, 0x00, 0x60, 0x04, 0x00};
+    dec3->data.assign(kEc3Data, kEc3Data + arraysize(kEc3Data));
+  }
+
   void Fill(AudioSampleEntry* enca) {
     enca->format = FOURCC_ENCA;
     enca->data_reference_index = 2;
@@ -895,6 +905,7 @@ class BoxDefinitionsTestGeneral : public testing::Test {
   bool IsOptional(const PixelAspectRatio* box) { return true; }
   bool IsOptional(const ElementaryStreamDescriptor* box) { return true; }
   bool IsOptional(const AC3Specific* box) { return true; }
+  bool IsOptional(const EC3Specific* box) { return true; }
   // Recommended, but optional.
   bool IsOptional(const WebVTTSourceLabelBox* box) { return true; }
   bool IsOptional(const CompositionTimeToSample* box) { return true; }
@@ -938,6 +949,7 @@ typedef testing::Types<FileType,
                        ElementaryStreamDescriptor,
                        DTSSpecific,
                        AC3Specific,
+                       EC3Specific,
                        AudioSampleEntry,
                        WebVTTConfigurationBox,
                        WebVTTSourceLabelBox,
@@ -1078,7 +1090,6 @@ TEST_F(BoxDefinitionsTest, DTSSampleEntry) {
   entry.samplerate = 44100;
   Fill(&entry.ddts);
   entry.Write(this->buffer_.get());
-
   AudioSampleEntry entry_readback;
   ASSERT_TRUE(ReadBack(&entry_readback));
   ASSERT_EQ(entry, entry_readback);
@@ -1099,6 +1110,21 @@ TEST_F(BoxDefinitionsTest, AC3SampleEntry) {
   ASSERT_EQ(entry, entry_readback);
 }
 
+TEST_F(BoxDefinitionsTest, EC3SampleEntry) {
+  AudioSampleEntry entry;
+  entry.format = FOURCC_EAC3;
+  entry.data_reference_index = 2;
+  entry.channelcount = 5;
+  entry.samplesize = 16;
+  entry.samplerate = 44100;
+  Fill(&entry.dec3);
+  entry.Write(this->buffer_.get());
+
+  AudioSampleEntry entry_readback;
+  ASSERT_TRUE(ReadBack(&entry_readback));
+  ASSERT_EQ(entry, entry_readback);
+}
+
 TEST_F(BoxDefinitionsTest, ProtectionSystemSpecificHeader) {
   ProtectionSystemSpecificHeader pssh;
   Fill(&pssh);

packager/media/formats/mp4/fourccs.h

@@ -26,6 +26,7 @@ enum FourCC {
   FOURCC_DAC3 = 0x64616333,
   FOURCC_DASH = 0x64617368,
   FOURCC_DDTS = 0x64647473,
+  FOURCC_DEC3 = 0x64656333,
   FOURCC_DINF = 0x64696e66,
   FOURCC_DREF = 0x64726566,
   FOURCC_DTSC = 0x64747363,
@@ -34,6 +35,7 @@ enum FourCC {
   FOURCC_DTSL = 0x6474736c,
   FOURCC_DTSM = 0x6474732d,
   FOURCC_DTSP = 0x6474732b,
+  FOURCC_EAC3 = 0x65632d33,  // This fourcc is "ec-3".
   FOURCC_EDTS = 0x65647473,
   FOURCC_ELST = 0x656c7374,
   FOURCC_ENCA = 0x656e6361,

packager/media/formats/mp4/mp4_media_parser.cc

@@ -76,6 +76,8 @@ AudioCodec FourCCToAudioCodec(FourCC fourcc) {
       return kCodecDTSM;
     case FOURCC_AC3:
       return kCodecAC3;
+    case FOURCC_EAC3:
+      return kCodecEAC3;
     default:
       return kUnknownAudioCodec;
   }
@@ -415,6 +417,11 @@ bool MP4MediaParser::ParseMoov(BoxReader* reader) {
           num_channels = entry.channelcount;
           sampling_frequency = entry.samplerate;
           break;
+        case FOURCC_EAC3:
+          extra_data = entry.dec3.data;
+          num_channels = entry.channelcount;
+          sampling_frequency = entry.samplerate;
+          break;
         default:
           LOG(ERROR) << "Unsupported audio format 0x" << std::hex
                      << actual_format << " in stsd box.";

packager/media/formats/mp4/mp4_muxer.cc

@@ -75,6 +75,8 @@ FourCC AudioCodecToFourCC(AudioCodec codec) {
       return FOURCC_DTSE;
     case kCodecDTSM:
       return FOURCC_DTSM;
+    case kCodecEAC3:
+      return FOURCC_EAC3;
     default:
       return FOURCC_NULL;
   }
@@ -266,6 +268,9 @@ void MP4Muxer::GenerateAudioTrak(const AudioStreamInfo* audio_info,
     case kCodecAC3:
       audio.dac3.data = audio_info->extra_data();
       break;
+    case kCodecEAC3:
+      audio.dec3.data = audio_info->extra_data();
+      break;
     default:
       NOTIMPLEMENTED();
       break;