iamf/cli/proto/mix_presentation.proto

// Copyright (c) 2023, Alliance for Open Media. All rights reserved
//
// This source code is subject to the terms of the BSD 3-Clause Clear License
// and the Alliance for Open Media Patent License 1.0. If the BSD 3-Clause Clear
// License was not distributed with this source code in the LICENSE file, you
// can obtain it at www.aomedia.org/license/software-license/bsd-3-c-c. If the
// Alliance for Open Media Patent License 1.0 was not distributed with this
// source code in the PATENTS file, you can obtain it at
// www.aomedia.org/license/patent.

syntax = "proto2";

package iamf_tools_cli_proto;

import "iamf/cli/proto/obu_header.proto";
import "iamf/cli/proto/param_definitions.proto";

message MixPresentationAnnotations {
  option deprecated = true;

  optional string mix_presentation_friendly_label = 1;
}

message MixPresentationElementAnnotations {
  option deprecated = true;

  optional string audio_element_friendly_label = 1;
}

message ElementMixConfig {
  option deprecated = true;

  optional MixGainParamDefinition mix_gain = 1;
}

enum HeadPhonesRenderingMode {
  HEADPHONES_RENDERING_MODE_INVALID = 0;
  HEADPHONES_RENDERING_MODE_STEREO = 1;
  HEADPHONES_RENDERING_MODE_BINAURAL = 2;
  HEADPHONES_RENDERING_MODE_RESERVED_2 = 3;
  HEADPHONES_RENDERING_MODE_RESERVED_3 = 4;
}

message RenderingConfig {
  optional HeadPhonesRenderingMode headphones_rendering_mode = 1;
  optional uint32 reserved = 2;
  optional uint32 rendering_config_extension_size = 3;
  optional bytes rendering_config_extension_bytes = 4;
}

message SubMixAudioElement {
  reserved 2;

  optional uint32 audio_element_id = 1;
  // Superseded by `localized_element_annotations`.
  repeated MixPresentationElementAnnotations
      mix_presentation_element_annotations_array = 4 [deprecated = true];
  // Length should be equal to `count_label` in the OBU.
  repeated string localized_element_annotations = 6;

  optional RenderingConfig rendering_config = 5;
  // Superseded by `element_mix_gain`.
  optional ElementMixConfig element_mix_config = 3 [deprecated = true];
  optional MixGainParamDefinition element_mix_gain = 7;
}

message OutputMixConfig {
  option deprecated = true;

  optional MixGainParamDefinition output_mix_gain = 1;
}

enum SoundSystem {
  SOUND_SYSTEM_INVALID = 0;
  SOUND_SYSTEM_A_0_2_0 = 1;
  SOUND_SYSTEM_B_0_5_0 = 2;
  SOUND_SYSTEM_C_2_5_0 = 3;
  SOUND_SYSTEM_D_4_5_0 = 4;
  SOUND_SYSTEM_E_4_5_1 = 5;
  SOUND_SYSTEM_F_3_7_0 = 6;
  SOUND_SYSTEM_G_4_9_0 = 7;
  SOUND_SYSTEM_H_9_10_3 = 8;
  SOUND_SYSTEM_I_0_7_0 = 9;
  SOUND_SYSTEM_J_4_7_0 = 10;
  SOUND_SYSTEM_10_2_7_0 = 11;
  SOUND_SYSTEM_11_2_3_0 = 12;
  SOUND_SYSTEM_12_0_1_0 = 13;
  SOUND_SYSTEM_13_6_9_0 = 14;
}

message LoudspeakersSsConventionLayout {
  optional SoundSystem sound_system = 1;
  optional uint32 reserved = 2;
}

message LoudspeakersReservedOrBinauralLayout {
  optional uint32 reserved = 1;
}

enum LayoutType {
  LAYOUT_TYPE_INVALID = 0;
  LAYOUT_TYPE_RESERVED_0 = 1;
  LAYOUT_TYPE_RESERVED_1 = 2;
  LAYOUT_TYPE_LOUDSPEAKERS_SS_CONVENTION = 3;
  LAYOUT_TYPE_BINAURAL = 4;
}

message Layout {
  reserved 2;

  optional LayoutType layout_type = 1;
  oneof specific_layout {
    LoudspeakersSsConventionLayout ss_layout = 3;
    LoudspeakersReservedOrBinauralLayout reserved_or_binaural_layout = 4;
  }
}

enum AnchorType {
  ANCHOR_TYPE_INVALID = 0;
  ANCHOR_TYPE_UNKNOWN = 1;
  ANCHOR_TYPE_DIALOGUE = 2;
  ANCHOR_TYPE_ALBUM = 3;
}

// Values are represented to align with the IAMF spec.
//
// See
// https://aomediacodec.github.io/iamf/v1.1.0.html#obu-mixpresentation-loudness
// for further details.
//
// See detailed examples on Q7.8 format below in `LoudnessInfo`.
message AnchorElement {
  optional AnchorType anchor_element = 1;
  optional int32 anchored_loudness = 2;  // Q7.8.
}

message AnchoredLoudness {
  // `num_anchored_loudness` is ignored. The value in the bitstream is inferred
  // based on the number of `anchor_elements`.
  optional uint32 num_anchored_loudness = 1 [deprecated = true];
  repeated AnchorElement anchor_elements = 2;
}

enum LoudnessInfoTypeBitMask {
  LOUDNESS_INFO_TYPE_INVALID = 0;
  LOUDNESS_INFO_TYPE_TRUE_PEAK = 1;
  LOUDNESS_INFO_TYPE_ANCHORED_LOUDNESS = 2;
  LOUDNESS_INFO_TYPE_RESERVED_4 = 4;
  LOUDNESS_INFO_TYPE_RESERVED_8 = 8;
  LOUDNESS_INFO_TYPE_RESERVED_16 = 16;
  LOUDNESS_INFO_TYPE_RESERVED_32 = 32;
  LOUDNESS_INFO_TYPE_RESERVED_64 = 64;
  LOUDNESS_INFO_TYPE_RESERVED_128 = 128;
}

// Values are represented to align with the IAMF spec.
//
// See
// https://aomediacodec.github.io/iamf/v1.1.0.html#obu-mixpresentation-loudness
// for further details.
//
// To convert from dB to Q7.8, multiply by 256. Example:
//  - For -3 dB, set the proto value to 256 * -3 = -768.
//  - For +6 dB, set the proto value to 256 * 6 = 1536.
message LoudnessInfo {
  optional uint32 deprecated_info_type = 1 [deprecated = true];
  repeated LoudnessInfoTypeBitMask info_type_bit_masks = 8 [packed = true];
  optional int32 integrated_loudness = 2;  // Q7.8.
  optional int32 digital_peak = 3;         // Q7.8.
  optional int32 true_peak = 4;            // Q7.8.
  optional AnchoredLoudness anchored_loudness = 5;
  optional uint32 info_type_size = 6;
  optional bytes info_type_bytes = 7;

  // Next ID: 9
}

message MixPresentationLayout {
  optional Layout loudness_layout = 1;
  optional LoudnessInfo loudness = 2;
}

message MixPresentationSubMix {
  // `num_audio_elements` is ignored. The value in the bitstream is inferred
  // based on the number of `audio_elements`.
  optional uint32 num_audio_elements = 1 [deprecated = true];
  repeated SubMixAudioElement audio_elements = 2;

  // Superseded by `output_mix_gain`.
  optional OutputMixConfig output_mix_config = 3 [deprecated = true];
  optional MixGainParamDefinition output_mix_gain = 6;

  // `num_layouts` is ignored. The value in the bitstream is inferred
  // based on the number of `layouts`.
  optional uint32 num_layouts = 4 [deprecated = true];
  repeated MixPresentationLayout layouts = 5;

  // Next ID: 7
}

message MixPresentationTag {
  optional string tag_name = 1;
  optional string tag_value = 2;
}

message MixPresentationTags {
  // `num_tags` is ignored. The value in the bitstream is inferred based on the
  // number of tags.
  optional uint32 num_tags = 1 [deprecated = true];
  repeated MixPresentationTag tags = 2;
}

message MixPresentationObuMetadata {
  reserved 2;

  optional uint32 mix_presentation_id = 1;
  optional uint32 count_label = 5;

  // Superseded by `annotations_language`.
  repeated string language_labels = 6 [deprecated = true];
  // Length should be equal to `count_label`.
  repeated string annotations_language = 9;

  // Superseded by `localized_presentation_annotations`.
  repeated MixPresentationAnnotations mix_presentation_annotations_array = 7
      [deprecated = true];
  // Length should be equal to `count_label`.
  repeated string localized_presentation_annotations = 10;

  // `num_sub_mixes` is ignored. The value in the bitstream is inferred
  // based on the number of `sub_mixes`.
  optional uint32 num_sub_mixes = 3 [deprecated = true];
  repeated MixPresentationSubMix sub_mixes = 4;

  // When false [default]: The encoder will ignore the below
  //                      `mix_presentation_tags`. Beware that other features
  //                      may still result in additional non-user specified
  //                      tags. To stricly encode a file without any tags, set
  //                      both `include_mix_presentation_tags` and
  //                      `EncoderControlMetadata.add_build_information_tag`
  //                      fields to false.
  // When true: The encoder will use the `mix_presentation_tags` field.
  optional bool include_mix_presentation_tags = 11;
  optional MixPresentationTags mix_presentation_tags = 12;

  optional ObuHeaderMetadata obu_header = 8;

  // Next ID: 13
}