Finalization of the first version of EchoCanceller 3

webrtc/modules/audio_processing/aec3/fft_buffer.cc

Index: webrtc/modules/audio_processing/aec3/fft_buffer.cc
diff --git a/webrtc/modules/audio_processing/aec3/fft_buffer.cc b/webrtc/modules/audio_processing/aec3/fft_buffer.cc
new file mode 100644
index 0000000000000000000000000000000000000000..6542d108ef99d9c363ef715af2ace132c03b656b
--- /dev/null
+++ b/webrtc/modules/audio_processing/aec3/fft_buffer.cc
@@ -0,0 +1,72 @@
+/*
+ *  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 "webrtc/modules/audio_processing/aec3/fft_buffer.h"
+
+#include <algorithm>
+
+#include "webrtc/base/checks.h"
+#include "webrtc/modules/audio_processing/aec3/aec3_common.h"
+
+namespace webrtc {
+
+FftBuffer::FftBuffer(Aec3Optimization optimization,
+                     size_t num_partitions,
+                     const std::vector<size_t> num_ffts_for_spectral_sums)
+    : optimization_(optimization),
+      fft_buffer_(num_partitions),
+      spectrum_buffer_(num_partitions, std::array<float, kFftLengthBy2Plus1>()),
+      spectral_sums_(num_ffts_for_spectral_sums.size(),
+                     std::array<float, kFftLengthBy2Plus1>()) {
+  // Current implementation only allows a maximum of one spectral sum lengths.
+  RTC_DCHECK_EQ(1, num_ffts_for_spectral_sums.size());
+  spectral_sums_length_ = num_ffts_for_spectral_sums[0];
+  RTC_DCHECK_GE(fft_buffer_.size(), spectral_sums_length_);
+
+  for (auto& sum : spectral_sums_) {
+    sum.fill(0.f);
+  }
+
+  for (auto& spectrum : spectrum_buffer_) {
+    spectrum.fill(0.f);
+  }
+
+  for (auto& fft : fft_buffer_) {
+    fft.Clear();
+  }
+}
+
+FftBuffer::~FftBuffer() = default;
+
+void FftBuffer::Insert(const FftData& fft) {
+  // Insert the fft into the buffer.
+  position_ = (position_ - 1 + fft_buffer_.size()) % fft_buffer_.size();
+  fft_buffer_[position_].Assign(fft);
+
+  // Compute and insert the spectrum for the FFT into the spectrum buffer.
+  fft.Spectrum(optimization_, &spectrum_buffer_[position_]);
+
+  // Pre-compute and cachec the spectral sums.

[aleloi, 2017/02/23 15:16:00]

cachec -> cache

+  std::copy(spectrum_buffer_[position_].begin(),
+            spectrum_buffer_[position_].end(), spectral_sums_[0].begin());

[aleloi, 2017/02/23 15:16:00]

Does this caching really save time? I profiled something similar recently where
3 frames were added (instead of 12), and there was no noticeable improvement.
Code would be a little simpler if all calculation is done in the loop.

+  size_t position = (position_ + 1) % fft_buffer_.size();
+  for (size_t j = 1; j < spectral_sums_length_; ++j) {
+    const std::array<float, kFftLengthBy2Plus1>& spectrum =
+        spectrum_buffer_[position];
+

[aleloi, 2017/02/23 15:16:00]

IMO readability improvement (optional): compute the position by

position = position_;
for j=1..:
  position = (condition) ? position + 1 : 0;
  // spectrum update


Alternatively update the position inside the loop header:

for (position = ..., j=...; condition; ++j, position = ...)

pro: scope of 'position' limited to loop.
con: more syntactic overhead.

+    for (size_t k = 0; k < spectral_sums_[0].size(); ++k) {
+      spectral_sums_[0][k] += spectrum[k];
+    }
+
+    position = position < (fft_buffer_.size() - 1) ? position + 1 : 0;
+  }
+}
+
+}  // namespace webrtc