Create an abstract RendererFactory.

PiperOrigin-RevId: 628081076

iamf/cli/BUILD

@@ -495,6 +495,19 @@ cc_library(
     ],
 )
 
+cc_library(
+    name = "renderer_factory",
+    srcs = ["renderer_factory.cc"],
+    hdrs = ["renderer_factory.h"],
+    deps = [
+        ":audio_element_with_data",
+        "//iamf/cli/renderer:audio_element_renderer_base",
+        "//iamf/obu:audio_element",
+        "//iamf/obu:leb128",
+        "//iamf/obu:mix_presentation",
+    ],
+)
+
 cc_library(
     name = "wav_reader",
     srcs = ["wav_reader.cc"],

iamf/cli/mix_presentation_finalizer.cc

@@ -117,7 +117,7 @@ MeasureLoudnessOrFallbackToUserLoudnessMixPresentationFinalizer::Finalize(
   LOG(INFO) << "  Loudness information may be copied from user "
             << "provided values.";
 
-  // TODO(b/332567539): Render and measure loudness for certain layouts.
+  // TODO(b/332567539): Use `RendererFactory` to render certain layouts.
   int metadata_index = 0;
   for (auto& mix_presentation_obu : mix_presentation_obus) {
     for (int sub_mix_index = 0;

iamf/cli/renderer_factory.cc

@@ -0,0 +1,40 @@
+/*
+ * Copyright (c) 2024, 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.
+ */
+#include "iamf/cli/renderer_factory.h"
+
+#include <memory>
+#include <vector>
+
+#include "iamf/cli/audio_element_with_data.h"
+#include "iamf/cli/renderer/audio_element_renderer_base.h"
+#include "iamf/obu/audio_element.h"
+#include "iamf/obu/leb128.h"
+#include "iamf/obu/mix_presentation.h"
+
+namespace iamf_tools {
+
+RendererFactoryBase::~RendererFactoryBase() {}
+
+std::unique_ptr<AudioElementRendererBase>
+RendererFactory::CreateRendererForLayout(
+    const std::vector<DecodedUleb128>& audio_substream_ids,
+    const SubstreamIdLabelsMap& substream_id_to_labels,
+    AudioElementObu::AudioElementType audio_element_type,
+    const AudioElementObu::AudioElementConfig& config,
+    const Layout& loudness_layout) const {
+  // TODO(b/332567539): Implement and return renderers depending on the input
+  //                    audio element and output layout.
+
+  return nullptr;
+}
+
+}  // namespace iamf_tools

iamf/cli/renderer_factory.h

@@ -0,0 +1,89 @@
+/*
+ * Copyright (c) 2024, 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.
+ */
+#ifndef CLI_RENDERER_FACTORY_H_
+#define CLI_RENDERER_FACTORY_H_
+
+#include <memory>
+#include <vector>
+
+#include "iamf/cli/audio_element_with_data.h"
+#include "iamf/cli/renderer/audio_element_renderer_base.h"
+#include "iamf/obu/audio_element.h"
+#include "iamf/obu/leb128.h"
+#include "iamf/obu/mix_presentation.h"
+
+namespace iamf_tools {
+
+/*\!brief Abstract class to create renderers.
+ *
+ * This class will be used when rendering the loudness of a mix presentation
+ * layout. The mix presentation finalizer will take in a factory and use them to
+ * create a renderers. By taking in a factory the finalizer can be agnostic to
+ * the collection of renderers that are being used and it what circumstances
+ * they are used.
+ */
+class RendererFactoryBase {
+ public:
+  /*\!brief Creates a renderer based on the audio element and layout.
+   *
+   * \param audio_substream_ids Audio susbtream IDs.
+   * \param substream_id_to_labels Mapping of substream IDs to labels.
+   * \param audio_element_type Type of the audio element.
+   * \param config Configuration of the audio element.
+   * \param loudness_layout Layout to render to.
+   * \return Unique pointer to an audio element renderer or `nullptr` if it not
+   *     known how to render the audio element.
+   */
+  virtual std::unique_ptr<AudioElementRendererBase> CreateRendererForLayout(
+      const std::vector<DecodedUleb128>& audio_substream_ids,
+      const SubstreamIdLabelsMap& substream_id_to_labels,
+      AudioElementObu::AudioElementType audio_element_type,
+      const AudioElementObu::AudioElementConfig& config,
+      const Layout& loudness_layout) const = 0;
+
+  /*\!brief Destructor. */
+  virtual ~RendererFactoryBase() = 0;
+};
+
+/*\!brief Factory which creates a renderers.
+ *
+ * This factory provides renderers in a best-effort manner according to the
+ * recommendations in the IAMF specification. When a recommended renderer is not
+ * implemented by `iamf-tools` the factory will fallback to returning a
+ * `nullptr`.
+ */
+class RendererFactory : public RendererFactoryBase {
+ public:
+  /*\!brief Creates a renderer based on the audio element and layout.
+   *
+   * \param audio_substream_ids Audio susbtream IDs.
+   * \param substream_id_to_labels Mapping of substream IDs to labels.
+   * \param audio_element_type Type of the audio element.
+   * \param config Configuration of the audio element.
+   * \param loudness_layout Layout to render to.
+   * \return Unique pointer to an audio element renderer or `nullptr` if it not
+   *     known how to render the audio element.
+   */
+  std::unique_ptr<AudioElementRendererBase> CreateRendererForLayout(
+      const std::vector<DecodedUleb128>& audio_substream_ids,
+      const SubstreamIdLabelsMap& substream_id_to_labels,
+      AudioElementObu::AudioElementType audio_element_type,
+      const AudioElementObu::AudioElementConfig& config,
+      const Layout& loudness_layout) const override;
+
+  /*\!brief Destructor. */
+  ~RendererFactory() override = default;
+};
+
+}  // namespace iamf_tools
+
+#endif  // CLI_LOUDNESS_CALCULATOR_H_

iamf/cli/tests/BUILD

@@ -377,6 +377,21 @@ cc_test(
     ],
 )
 
+cc_test(
+    name = "renderer_factory_test",
+    srcs = ["renderer_factory_test.cc"],
+    deps = [
+        "//iamf/cli:renderer_factory",
+        "//iamf/cli/proto:obu_header_cc_proto",
+        "//iamf/cli/proto:parameter_data_cc_proto",
+        "//iamf/cli/proto:temporal_delimiter_cc_proto",
+        "//iamf/cli/proto:user_metadata_cc_proto",
+        "//iamf/obu:audio_element",
+        "//iamf/obu:mix_presentation",
+        "@com_google_googletest//:gtest_main",
+    ],
+)
+
 cc_test(
     name = "wav_reader_test",
     srcs = ["wav_reader_test.cc"],

iamf/cli/tests/renderer_factory_test.cc

@@ -0,0 +1,95 @@
+/*
+ * Copyright (c) 2024, 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.
+ */
+#include "iamf/cli/renderer_factory.h"
+
+#include "gtest/gtest.h"
+#include "iamf/cli/proto/obu_header.pb.h"
+#include "iamf/cli/proto/parameter_data.pb.h"
+#include "iamf/cli/proto/temporal_delimiter.pb.h"
+#include "iamf/cli/proto/user_metadata.pb.h"
+#include "iamf/obu/audio_element.h"
+#include "iamf/obu/mix_presentation.h"
+
+namespace iamf_tools {
+namespace {
+
+using enum LoudspeakersSsConventionLayout::SoundSystem;
+using enum ChannelAudioLayerConfig::LoudspeakerLayout;
+
+const Layout kMonoLayout = {
+    .layout_type = Layout::kLayoutTypeLoudspeakersSsConvention,
+    .specific_layout =
+        LoudspeakersSsConventionLayout{.sound_system = kSoundSystem12_0_1_0}};
+const Layout kBinauralLayout = {.layout_type = Layout::kLayoutTypeBinaural};
+
+const ScalableChannelLayoutConfig kMonoScalableChannelLayoutConfig = {
+    .num_layers = 1,
+    .channel_audio_layer_configs = {{.loudspeaker_layout = kLayoutMono}}};
+
+const ScalableChannelLayoutConfig kStereoScalableChannelLayoutConfig = {
+    .num_layers = 1,
+    .channel_audio_layer_configs = {{.loudspeaker_layout = kLayoutStereo}}};
+
+const AmbisonicsConfig kFullZerothOrderAmbisonicsConfig = {
+    .ambisonics_mode = AmbisonicsConfig::kAmbisonicsModeMono,
+    .ambisonics_config = AmbisonicsMonoConfig{.output_channel_count = 1,
+                                              .substream_count = 1,
+                                              .channel_mapping = {0}}};
+
+TEST(CreateRendererForLayout, ReturnsNullPtrForPassThroughRenderer) {
+  const RendererFactory factory;
+
+  EXPECT_EQ(factory.CreateRendererForLayout(
+                {0}, {{0, {"M"}}}, AudioElementObu::kAudioElementChannelBased,
+                kMonoScalableChannelLayoutConfig, kMonoLayout),
+            nullptr);
+}
+
+TEST(CreateRendererForLayout, ReturnsNullPtrForChannelToBinauralRenderer) {
+  const RendererFactory factory;
+
+  EXPECT_EQ(factory.CreateRendererForLayout(
+                {0}, {{0, {"M"}}}, AudioElementObu::kAudioElementChannelBased,
+                kMonoScalableChannelLayoutConfig, kBinauralLayout),
+            nullptr);
+}
+
+TEST(CreateRendererForLayout, ReturnsNullPtrForChannelToChannelRenderer) {
+  const RendererFactory factory;
+
+  EXPECT_EQ(
+      factory.CreateRendererForLayout(
+          {0}, {{0, {"L2", "R2"}}}, AudioElementObu::kAudioElementChannelBased,
+          kStereoScalableChannelLayoutConfig, kMonoLayout),
+      nullptr);
+}
+
+TEST(CreateRendererForLayout, ReturnsNullPtrForAmbisonicsToChannelRenderer) {
+  const RendererFactory factory;
+
+  EXPECT_EQ(factory.CreateRendererForLayout(
+                {0}, {{0, {"A0"}}}, AudioElementObu::kAudioElementSceneBased,
+                kFullZerothOrderAmbisonicsConfig, kMonoLayout),
+            nullptr);
+}
+
+TEST(CreateRendererForLayout, ReturnsNullPtrForAmbisonicsToBinauralRenderer) {
+  const RendererFactory factory;
+
+  EXPECT_EQ(factory.CreateRendererForLayout(
+                {0}, {{0, {"A0"}}}, AudioElementObu::kAudioElementSceneBased,
+                kFullZerothOrderAmbisonicsConfig, kBinauralLayout),
+            nullptr);
+}
+
+}  // namespace
+}  // namespace iamf_tools

iamf/obu/audio_element.h

@@ -217,6 +217,10 @@ class AudioElementObu : public ObuBase {
     kAudioElementEndReserved = 7,
   };
 
+  typedef std::variant<ScalableChannelLayoutConfig, AmbisonicsConfig,
+                       ExtensionConfig>
+      AudioElementConfig;
+
   /*!\brief Constructor.
    *
    * \param header `ObuHeader` of the OBU.
@@ -332,8 +336,7 @@ class AudioElementObu : public ObuBase {
   std::vector<AudioElementParam> audio_element_params_;
 
   // Active field depends on `audio_element_type_`.
-  std::variant<ScalableChannelLayoutConfig, AmbisonicsConfig, ExtensionConfig>
-      config_;
+  AudioElementConfig config_;
 
  private:
   DecodedUleb128 audio_element_id_;