diff --git a/modules/audio_processing/aec3/echo_canceller3.cc b/modules/audio_processing/aec3/echo_canceller3.cc
index e387940885..8c8f8bbd2b 100644
--- a/modules/audio_processing/aec3/echo_canceller3.cc
+++ b/modules/audio_processing/aec3/echo_canceller3.cc
@@ -445,6 +445,25 @@ bool EchoCanceller3::ActiveProcessing() const {
   return true;
 }
 
+EchoCanceller3Config EchoCanceller3::CreateDefaultConfig(
+    size_t num_render_channels,
+    size_t num_capture_channels) {
+  EchoCanceller3Config cfg;
+  if (num_render_channels > 1) {
+    // Use shorter and more rapidly adapting shadow filter to compensate for
+    // thge increased number of total filter parameters to adapt.
+    cfg.filter.shadow.length_blocks = 11;
+    cfg.filter.shadow.rate = 0.95f;
+    cfg.filter.shadow_initial.length_blocks = 11;
+    cfg.filter.shadow_initial.rate = 0.95f;
+
+    // Use more concervative suppressor behavior for non-nearend speech.
+    cfg.suppressor.normal_tuning.max_dec_factor_lf = 0.35f;
+    cfg.suppressor.normal_tuning.max_inc_factor = 1.5f;
+  }
+  return cfg;
+}
+
 void EchoCanceller3::EmptyRenderQueue() {
   RTC_DCHECK_RUNS_SERIALIZED(&capture_race_checker_);
   bool frame_to_buffer =
diff --git a/modules/audio_processing/aec3/echo_canceller3.h b/modules/audio_processing/aec3/echo_canceller3.h
index a828d5712b..f346f189bf 100644
--- a/modules/audio_processing/aec3/echo_canceller3.h
+++ b/modules/audio_processing/aec3/echo_canceller3.h
@@ -124,6 +124,11 @@ class EchoCanceller3 : public EchoControl {
     block_processor_->UpdateEchoLeakageStatus(leakage_detected);
   }
 
+  // Produces a default configuration that is suitable for a certain combination
+  // of render and capture channels.
+  static EchoCanceller3Config CreateDefaultConfig(size_t num_render_channels,
+                                                  size_t num_capture_channels);
+
  private:
   class RenderWriter;
 
diff --git a/modules/audio_processing/audio_processing_impl.cc b/modules/audio_processing/audio_processing_impl.cc
index 0e375c9b9b..3e49c0cb76 100644
--- a/modules/audio_processing/audio_processing_impl.cc
+++ b/modules/audio_processing/audio_processing_impl.cc
@@ -75,6 +75,14 @@ bool DetectLegacyNsEnforcement() {
   return field_trial::IsEnabled("WebRTC-NewNoiseSuppressionKillSwitch");
 }
 
+// Checks whether AEC3 should be allowed to decide what the default
+// configuration should be based on the render and capture channel configuration
+// at hand.
+bool UseSetupSpecificDefaultAec3Congfig() {
+  return !field_trial::IsEnabled(
+      "WebRTC-Aec3SetupSpecificDefaultConfigDefaultsKillSwitch");
+}
+
 // Identify the native processing rate that best handles a sample rate.
 int SuitableProcessRate(int minimum_rate,
                         int max_splitting_rate,
@@ -312,6 +320,8 @@ AudioProcessingImpl::AudioProcessingImpl(
     : data_dumper_(
           new ApmDataDumper(rtc::AtomicOps::Increment(&instance_count_))),
       enforced_usage_of_legacy_ns_(DetectLegacyNsEnforcement()),
+      use_setup_specific_default_aec3_config_(
+          UseSetupSpecificDefaultAec3Congfig()),
       capture_runtime_settings_(kRuntimeSettingQueueSize),
       render_runtime_settings_(kRuntimeSettingQueueSize),
       capture_runtime_settings_enqueuer_(&capture_runtime_settings_),
@@ -1826,8 +1836,13 @@ void AudioProcessingImpl::InitializeEchoController() {
           proc_sample_rate_hz(), num_reverse_channels(), num_proc_channels());
       RTC_DCHECK(submodules_.echo_controller);
     } else {
+      EchoCanceller3Config config =
+          use_setup_specific_default_aec3_config_
+              ? EchoCanceller3::CreateDefaultConfig(num_reverse_channels(),
+                                                    num_proc_channels())
+              : EchoCanceller3Config();
       submodules_.echo_controller = std::make_unique<EchoCanceller3>(
-          EchoCanceller3Config(), proc_sample_rate_hz(), num_reverse_channels(),
+          config, proc_sample_rate_hz(), num_reverse_channels(),
           num_proc_channels());
     }
 
diff --git a/modules/audio_processing/audio_processing_impl.h b/modules/audio_processing/audio_processing_impl.h
index 5f1d12ff32..e5d0573e12 100644
--- a/modules/audio_processing/audio_processing_impl.h
+++ b/modules/audio_processing/audio_processing_impl.h
@@ -153,6 +153,7 @@ class AudioProcessingImpl : public AudioProcessing {
   std::unique_ptr<ApmDataDumper> data_dumper_;
   static int instance_count_;
   const bool enforced_usage_of_legacy_ns_;
+  const bool use_setup_specific_default_aec3_config_;
 
   SwapQueue<RuntimeSetting> capture_runtime_settings_;
   SwapQueue<RuntimeSetting> render_runtime_settings_;