feat: Add xHE-AAC support

CONTRIBUTORS

@@ -31,6 +31,7 @@ Daniel Cantarín <canta@canta.com.ar>
 David Cavar <pal3thorn@gmail.com>
 Evgeny Zajcev <zevlg@yandex.ru>
 Gabe Kopley <gabe@philo.com>
+Geoff Jukes <geoff@jukes.org>
 Haoming Chen <hmchen@google.com>
 Jacob Trimble <modmaker@google.com>
 Joe Foraci <jforaci@gmail.com>

packager/media/codecs/aac_audio_specific_config.cc

@@ -62,10 +62,9 @@ bool AACAudioSpecificConfig::Parse(const std::vector<uint8_t>& data) {
   // Syntax of AudioSpecificConfig.
 
   // Read base configuration.
-  // Audio Object Types specified in ISO 14496-3, Table 1.15.
-  RCHECK(reader.ReadBits(5, &audio_object_type_));
-  // Audio objects type >=31 is not supported yet.
-  RCHECK(audio_object_type_ < 31);
+  // Audio Object Types specified in "ISO/IEC 14496-3:2019, Table 1.19"
+  RCHECK(ParseAudioObjectType(&reader));
+
   RCHECK(reader.ReadBits(4, &frequency_index_));
   if (frequency_index_ == 0xf)
     RCHECK(reader.ReadBits(24, &frequency_));
@@ -82,9 +81,7 @@ bool AACAudioSpecificConfig::Parse(const std::vector<uint8_t>& data) {
     RCHECK(reader.ReadBits(4, &extension_frequency_index));
     if (extension_frequency_index == 0xf)
       RCHECK(reader.ReadBits(24, &extension_frequency_));
-    RCHECK(reader.ReadBits(5, &audio_object_type_));
-    // Audio objects type >=31 is not supported yet.
-    RCHECK(audio_object_type_ < 31);
+    RCHECK(ParseAudioObjectType(&reader));
   }
 
   RCHECK(ParseDecoderGASpecificConfig(&reader));
@@ -131,10 +128,14 @@ bool AACAudioSpecificConfig::Parse(const std::vector<uint8_t>& data) {
     RCHECK(extension_frequency_index < arraysize(kSampleRates));
     extension_frequency_ = kSampleRates[extension_frequency_index];
   }
-
-  return frequency_ != 0 && num_channels_ != 0 && audio_object_type_ >= 1 &&
-         audio_object_type_ <= 4 && frequency_index_ != 0xf &&
-         channel_config_ <= 7;
+
+  if (audio_object_type_ == AOT_USAC) {
+    return frequency_ != 0 && num_channels_ != 0 && channel_config_ <= 7;
+  } else {
+    return frequency_ != 0 && num_channels_ != 0 && audio_object_type_ >= 1 &&
+           audio_object_type_ <= 4 && frequency_index_ != 0xf &&
+           channel_config_ <= 7;
+  }
 }
 
 bool AACAudioSpecificConfig::ConvertToADTS(
@@ -205,6 +206,18 @@ uint8_t AACAudioSpecificConfig::GetNumChannels() const {
   return num_channels_;
 }
 
+bool AACAudioSpecificConfig::ParseAudioObjectType(BitReader* bit_reader) {
+  RCHECK(bit_reader->ReadBits(5, &audio_object_type_));
+
+  if (audio_object_type_ == AOT_ESCAPE) {
+    uint8_t audioObjectTypeExt;
+    RCHECK(bit_reader->ReadBits(6, &audioObjectTypeExt));
+    audio_object_type_ = static_cast<AudioObjectType>(32 + audioObjectTypeExt);
+  }
+
+  return true;
+}
+
 // Currently this function only support GASpecificConfig defined in
 // ISO 14496 Part 3 Table 4.1 - Syntax of GASpecificConfig()
 bool AACAudioSpecificConfig::ParseDecoderGASpecificConfig(
@@ -223,6 +236,10 @@ bool AACAudioSpecificConfig::ParseDecoderGASpecificConfig(
     case 22:
     case 23:
       return ParseGASpecificConfig(bit_reader);
+    case 42:
+      // Skip UsacConfig() parsing until required
+      RCHECK(bit_reader->SkipBits(bit_reader->bits_available()));
+      return true;
     default:
       break;
   }

packager/media/codecs/aac_audio_specific_config.h

@@ -66,10 +66,10 @@ class AACAudioSpecificConfig {
     AOT_ER_AAC_ELD       = 39,  // Error Resilient Enhanced Low Delay
     AOT_SMR_SIMPLE       = 40,  // Symbolic Music Representation Simple
     AOT_SMR_MAIN         = 41,  // Symbolic Music Representation Main
-    AOT_USAC_NOSBR       = 42,  // Unified Speech and Audio Coding (no SBR)
+    AOT_USAC             = 42,  // Unified Speech and Audio Coding
     AOT_SAOC             = 43,  // Spatial Audio Object Coding
     AOT_LD_SURROUND      = 44,  // Low Delay MPEG Surround
-    AOT_USAC             = 45,  // Unified Speech and Audio Coding
+    SAOC_DE              = 45,  // Spatial Audio Object Coding Dialogue Enhancement
   };
 
   AACAudioSpecificConfig();
@@ -114,6 +114,7 @@ class AACAudioSpecificConfig {
   void set_sbr_present(bool sbr_present) { sbr_present_ = sbr_present; }
 
  private:
+  bool ParseAudioObjectType(BitReader* bit_reader);
   bool ParseDecoderGASpecificConfig(BitReader* bit_reader);
   bool SkipErrorSpecificConfig() const;
   // Parse GASpecificConfig. Calls |ParseProgramConfigElement| if

packager/media/codecs/aac_audio_specific_config_unittest.cc

@@ -101,6 +101,23 @@ TEST(AACAudioSpecificConfigTest, SixChannelTest) {
             aac_audio_specific_config.GetAudioObjectType());
 }
 
+TEST(AACAudioSpecificConfigTest, UsacTest) {
+  AACAudioSpecificConfig aac_audio_specific_config;
+  uint8_t buffer[] = { 
+    0xF9, 0x5E, 0x01, 0x2C, 0x00, 0x52, 0x42, 0x2C, 0xC0, 0x51,
+    0x17, 0x55, 0x4F, 0x36, 0x00, 0x42, 0x80, 0x01, 0x00, 0x04,
+    0xA8, 0x82, 0x34, 0xE5, 0x80
+  };
+
+  std::vector<uint8_t> data(std::begin(buffer), std::end(buffer));
+
+  EXPECT_TRUE(aac_audio_specific_config.Parse(data));
+  EXPECT_EQ(38400u, aac_audio_specific_config.GetSamplesPerSecond());
+  EXPECT_EQ(2u, aac_audio_specific_config.GetNumChannels());
+  EXPECT_EQ(AACAudioSpecificConfig::AOT_USAC,
+            aac_audio_specific_config.GetAudioObjectType());
+}
+
 TEST(AACAudioSpecificConfigTest, ProgramConfigElementTest) {
   uint8_t buffer[] = {
       0x11, 0x80, 0x04, 0xC8, 0x44, 0x00, 0x20, 0x00, 0xC4,