AEC3: Adding the option for applying a fixed delay to the capture signal

This CL adds functionality for applying an optional fixed delay in AEC3
to the capture signal

Bug: webrtc:9647
Change-Id: Id3b3f896bcf203e6611298dc804c3c80da9f1883
Reviewed-on: https://webrtc-review.googlesource.com/95142
Commit-Queue: Per Åhgren <peah@webrtc.org>
Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#24399}

api/audio/echo_canceller3_config.cc

@@ -14,6 +14,10 @@ namespace webrtc {
 EchoCanceller3Config::EchoCanceller3Config() = default;
 EchoCanceller3Config::EchoCanceller3Config(const EchoCanceller3Config& e) =
     default;
+EchoCanceller3Config::Delay::Delay() = default;
+EchoCanceller3Config::Delay::Delay(const EchoCanceller3Config::Delay& e) =
+    default;
+
 EchoCanceller3Config::Mask::Mask() = default;
 EchoCanceller3Config::Mask::Mask(const EchoCanceller3Config::Mask& m) = default;
 

api/audio/echo_canceller3_config.h

@@ -20,6 +20,8 @@ struct EchoCanceller3Config {
   EchoCanceller3Config();
   EchoCanceller3Config(const EchoCanceller3Config& e);
   struct Delay {
+    Delay();
+    Delay(const Delay& e);
     size_t default_delay = 5;
     size_t down_sampling_factor = 4;
     size_t num_filters = 6;
@@ -29,6 +31,7 @@ struct EchoCanceller3Config {
     size_t hysteresis_limit_1_blocks = 1;
     size_t hysteresis_limit_2_blocks = 1;
     size_t skew_hysteresis_blocks = 3;
+    size_t fixed_capture_delay_samples = 0;
   } delay;
 
   struct Filter {

modules/audio_processing/aec3/BUILD.gn

@@ -20,6 +20,8 @@ rtc_static_library("aec3") {
     "aec3_fft.h",
     "aec_state.cc",
     "aec_state.h",
+    "block_delay_buffer.cc",
+    "block_delay_buffer.h",
     "block_framer.cc",
     "block_framer.h",
     "block_processor.cc",
@@ -179,6 +181,7 @@ if (rtc_include_tests) {
         "adaptive_fir_filter_unittest.cc",
         "aec3_fft_unittest.cc",
         "aec_state_unittest.cc",
+        "block_delay_buffer_unittest.cc",
         "block_framer_unittest.cc",
         "block_processor_metrics_unittest.cc",
         "block_processor_unittest.cc",

modules/audio_processing/aec3/block_delay_buffer.cc

@@ -0,0 +1,47 @@
+/*
+ *  Copyright (c) 2018 The WebRTC project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+#include "modules/audio_processing/aec3/block_delay_buffer.h"
+
+#include "rtc_base/checks.h"
+
+namespace webrtc {
+
+BlockDelayBuffer::BlockDelayBuffer(size_t num_bands,
+                                   size_t frame_length,
+                                   size_t delay_samples)
+    : frame_length_(frame_length),
+      delay_(delay_samples),
+      buf_(num_bands, std::vector<float>(delay_, 0.f)) {}
+
+BlockDelayBuffer::~BlockDelayBuffer() = default;
+
+void BlockDelayBuffer::DelaySignal(AudioBuffer* frame) {
+  RTC_DCHECK_EQ(1, frame->num_channels());
+  RTC_DCHECK_EQ(buf_.size(), frame->num_bands());
+  if (delay_ == 0) {
+    return;
+  }
+
+  const size_t i_start = last_insert_;
+  size_t i = 0;
+  for (size_t j = 0; j < buf_.size(); ++j) {
+    i = i_start;
+    for (size_t k = 0; k < frame_length_; ++k) {
+      const float tmp = buf_[j][i];
+      buf_[j][i] = frame->split_bands_f(0)[j][k];
+      frame->split_bands_f(0)[j][k] = tmp;
+      i = i < buf_[0].size() - 1 ? i + 1 : 0;
+    }
+  }
+
+  last_insert_ = i;
+}
+
+}  // namespace webrtc

modules/audio_processing/aec3/block_delay_buffer.h

@@ -0,0 +1,38 @@
+/*
+ *  Copyright (c) 2018 The WebRTC project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#ifndef MODULES_AUDIO_PROCESSING_AEC3_BLOCK_DELAY_BUFFER_H_
+#define MODULES_AUDIO_PROCESSING_AEC3_BLOCK_DELAY_BUFFER_H_
+
+#include <vector>
+
+#include "modules/audio_processing/audio_buffer.h"
+
+namespace webrtc {
+
+// Class for applying a fixed delay to the samples in a signal partitioned using
+// the audiobuffer band-splitting scheme.
+class BlockDelayBuffer {
+ public:
+  BlockDelayBuffer(size_t num_bands, size_t frame_length, size_t delay_samples);
+  ~BlockDelayBuffer();
+
+  // Delays the samples by the specified delay.
+  void DelaySignal(AudioBuffer* frame);
+
+ private:
+  const size_t frame_length_;
+  const size_t delay_;
+  std::vector<std::vector<float>> buf_;
+  size_t last_insert_ = 0;
+};
+}  // namespace webrtc
+
+#endif  // MODULES_AUDIO_PROCESSING_AEC3_BLOCK_DELAY_BUFFER_H_

modules/audio_processing/aec3/block_delay_buffer_unittest.cc

@@ -0,0 +1,91 @@
+/*
+ *  Copyright (c) 2017 The WebRTC project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "modules/audio_processing/aec3/block_delay_buffer.h"
+
+#include <string>
+
+#include "modules/audio_processing/aec3/aec3_common.h"
+#include "modules/audio_processing/audio_buffer.h"
+#include "rtc_base/strings/string_builder.h"
+#include "test/gtest.h"
+
+namespace webrtc {
+
+namespace {
+
+float SampleValue(size_t sample_index) {
+  return sample_index % 32768;
+}
+
+// Populates the frame with linearly increasing sample values for each band.
+void PopulateInputFrame(size_t frame_length,
+                        size_t num_bands,
+                        size_t first_sample_index,
+                        float* const* frame) {
+  for (size_t k = 0; k < num_bands; ++k) {
+    for (size_t i = 0; i < frame_length; ++i) {
+      frame[k][i] = SampleValue(first_sample_index + i);
+    }
+  }
+}
+
+std::string ProduceDebugText(int sample_rate_hz, size_t delay) {
+  char log_stream_buffer[8 * 1024];
+  rtc::SimpleStringBuilder ss(log_stream_buffer);
+  ss << "Sample rate: " << sample_rate_hz;
+  ss << ", Delay: " << delay;
+  return ss.str();
+}
+
+}  // namespace
+
+// Verifies that the correct signal delay is achived.
+TEST(BlockDelayBuffer, CorrectDelayApplied) {
+  for (size_t delay : {0, 1, 27, 160, 4321, 7021}) {
+    for (auto rate : {8000, 16000, 32000, 48000}) {
+      SCOPED_TRACE(ProduceDebugText(rate, delay));
+      size_t num_bands = NumBandsForRate(rate);
+      size_t fullband_frame_length = rate / 100;
+      size_t subband_frame_length = rate == 8000 ? 80 : 160;
+
+      BlockDelayBuffer delay_buffer(num_bands, subband_frame_length, delay);
+
+      static constexpr size_t kNumFramesToProcess = 20;
+      for (size_t frame_index = 0; frame_index < kNumFramesToProcess;
+           ++frame_index) {
+        AudioBuffer audio_buffer(fullband_frame_length, 1,
+                                 fullband_frame_length, 1,
+                                 fullband_frame_length);
+        if (rate > 16000) {
+          audio_buffer.SplitIntoFrequencyBands();
+        }
+        size_t first_sample_index = frame_index * subband_frame_length;
+        PopulateInputFrame(subband_frame_length, num_bands, first_sample_index,
+                           &audio_buffer.split_bands_f(0)[0]);
+        delay_buffer.DelaySignal(&audio_buffer);
+
+        for (size_t k = 0; k < num_bands; ++k) {
+          size_t sample_index = first_sample_index;
+          for (size_t i = 0; i < subband_frame_length; ++i, ++sample_index) {
+            if (sample_index < delay) {
+              EXPECT_EQ(0.f, audio_buffer.split_bands_f(0)[k][i]);
+            } else {
+              EXPECT_EQ(SampleValue(sample_index - delay),
+                        audio_buffer.split_bands_f(0)[k][i]);
+            }
+          }
+        }
+      }
+    }
+  }
+}
+
+}  // namespace webrtc

modules/audio_processing/aec3/echo_canceller3.cc

@@ -342,6 +342,7 @@ EchoCanceller3::EchoCanceller3(const EchoCanceller3Config& config,
                                std::unique_ptr<BlockProcessor> block_processor)
     : data_dumper_(
           new ApmDataDumper(rtc::AtomicOps::Increment(&instance_count_))),
+      config_(config),
       sample_rate_hz_(sample_rate_hz),
       num_bands_(NumBandsForRate(sample_rate_hz_)),
       frame_length_(rtc::CheckedDivExact(LowestBandRate(sample_rate_hz_), 100)),
@@ -358,7 +359,10 @@ EchoCanceller3::EchoCanceller3(const EchoCanceller3Config& config,
       render_queue_output_frame_(num_bands_,
                                  std::vector<float>(frame_length_, 0.f)),
       block_(num_bands_, std::vector<float>(kBlockSize, 0.f)),
-      sub_frame_view_(num_bands_) {
+      sub_frame_view_(num_bands_),
+      block_delay_buffer_(num_bands_,
+                          frame_length_,
+                          config_.delay.fixed_capture_delay_samples) {
   RTC_DCHECK(ValidFullBandRate(sample_rate_hz_));
 
   std::unique_ptr<CascadedBiQuadFilter> render_highpass_filter;
@@ -421,6 +425,11 @@ void EchoCanceller3::ProcessCapture(AudioBuffer* capture, bool level_change) {
   data_dumper_->DumpRaw("aec3_call_order",
                         static_cast<int>(EchoCanceller3ApiCall::kCapture));
 
+  // Optionally delay the capture signal.
+  if (config_.delay.fixed_capture_delay_samples > 0) {
+    block_delay_buffer_.DelaySignal(capture);
+  }
+
   rtc::ArrayView<float> capture_lower_band =
       rtc::ArrayView<float>(&capture->split_bands_f(0)[0][0], frame_length_);
 

modules/audio_processing/aec3/echo_canceller3.h

@@ -12,6 +12,7 @@
 #define MODULES_AUDIO_PROCESSING_AEC3_ECHO_CANCELLER3_H_
 
 #include "api/audio/echo_canceller3_config.h"
+#include "modules/audio_processing/aec3/block_delay_buffer.h"
 #include "modules/audio_processing/aec3/block_framer.h"
 #include "modules/audio_processing/aec3/block_processor.h"
 #include "modules/audio_processing/aec3/cascaded_biquad_filter.h"
@@ -110,6 +111,7 @@ class EchoCanceller3 : public EchoControl {
   // State that may be accessed by the capture thread.
   static int instance_count_;
   std::unique_ptr<ApmDataDumper> data_dumper_;
+  const EchoCanceller3Config config_;
   const int sample_rate_hz_;
   const int num_bands_;
   const size_t frame_length_;
@@ -129,6 +131,7 @@ class EchoCanceller3 : public EchoControl {
   std::vector<std::vector<float>> block_ RTC_GUARDED_BY(capture_race_checker_);
   std::vector<rtc::ArrayView<float>> sub_frame_view_
       RTC_GUARDED_BY(capture_race_checker_);
+  BlockDelayBuffer block_delay_buffer_ RTC_GUARDED_BY(capture_race_checker_);
 
   RTC_DISALLOW_IMPLICIT_CONSTRUCTORS(EchoCanceller3);
 };

modules/audio_processing/test/audio_processing_simulator.cc

@@ -203,6 +203,8 @@ EchoCanceller3Config ParseAec3Parameters(const std::string& filename) {
               &cfg.delay.hysteresis_limit_2_blocks);
     ReadParam(section, "skew_hysteresis_blocks",
               &cfg.delay.skew_hysteresis_blocks);
+    ReadParam(section, "fixed_capture_delay_samples",
+              &cfg.delay.fixed_capture_delay_samples);
   }
 
   if (rtc::GetValueFromJsonObject(root, "filter", &section)) {