Reland "RNN VAD: Replace Ooura with PFFFT for the pitch auto correlation."
This reverts commit d38ce9f82402a995289b87e044125847d12eb658.
Reason for revert: land this first, then fix downstream
Original change's description:
> Revert "RNN VAD: Replace Ooura with PFFFT for the pitch auto correlation."
>
> This reverts commit 8fcd6537f242ffd74154a62dad410e573e2efc4b.
>
> Reason for revert: broke internal projects.
>
> Original change's description:
> > RNN VAD: Replace Ooura with PFFFT for the pitch auto correlation.
> >
> > Bug: webrtc:9577, webrtc:10480
> > Change-Id: I6d58866d48b8eaaa4102551b88d4f55133d1915c
> > Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/130482
> > Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
> > Reviewed-by: Gustaf Ullberg <gustaf@webrtc.org>
> > Cr-Commit-Position: refs/heads/master@{#27387}
>
> TBR=gustaf@webrtc.org,alessiob@webrtc.org
>
> Change-Id: Ia05057326ebc277f334b13db0bfec9d4442903c2
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: webrtc:9577, webrtc:10480
> Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/130369
> Reviewed-by: Qingsi Wang <qingsi@webrtc.org>
> Commit-Queue: Qingsi Wang <qingsi@webrtc.org>
> Cr-Commit-Position: refs/heads/master@{#27405}
TBR=gustaf@webrtc.org,alessiob@webrtc.org,qingsi@webrtc.org
Change-Id: I345f00d798fb67d0f02bf1e89319dac7f0287485
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: webrtc:9577, webrtc:10480
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/130516
Reviewed-by: Alessio Bazzica <alessiob@webrtc.org>
Commit-Queue: Alessio Bazzica <alessiob@webrtc.org>
Cr-Commit-Position: refs/heads/master@{#27407}diff --git a/modules/audio_processing/agc2/rnn_vad/BUILD.gn b/modules/audio_processing/agc2/rnn_vad/BUILD.gn
index 7379d41..237c809 100644
--- a/modules/audio_processing/agc2/rnn_vad/BUILD.gn
+++ b/modules/audio_processing/agc2/rnn_vad/BUILD.gn
@@ -11,6 +11,8 @@
rtc_source_set("rnn_vad") {
visibility = [ "../*" ]
sources = [
+ "auto_correlation.cc",
+ "auto_correlation.h",
"common.h",
"features_extraction.cc",
"features_extraction.h",
@@ -37,9 +39,9 @@
"..:biquad_filter",
"../../../../api:array_view",
"../../../../api:function_view",
- "../../../../common_audio/",
"../../../../rtc_base:checks",
"../../../../rtc_base:rtc_base_approved",
+ "../../utility:pffft_wrapper",
"//third_party/rnnoise:kiss_fft",
"//third_party/rnnoise:rnn_vad",
]
@@ -53,6 +55,7 @@
"test_utils.h",
]
deps = [
+ ":rnn_vad",
"../../../../api:array_view",
"../../../../api:scoped_refptr",
"../../../../rtc_base:checks",
@@ -86,6 +89,7 @@
rtc_source_set("unittests") {
testonly = true
sources = [
+ "auto_correlation_unittest.cc",
"features_extraction_unittest.cc",
"fft_util_unittest.cc",
"lp_residual_unittest.cc",
diff --git a/modules/audio_processing/agc2/rnn_vad/auto_correlation.cc b/modules/audio_processing/agc2/rnn_vad/auto_correlation.cc
new file mode 100644
index 0000000..d932c78
--- /dev/null
+++ b/modules/audio_processing/agc2/rnn_vad/auto_correlation.cc
@@ -0,0 +1,92 @@
+/*
+ * Copyright (c) 2019 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/agc2/rnn_vad/auto_correlation.h"
+
+#include <algorithm>
+
+#include "rtc_base/checks.h"
+
+namespace webrtc {
+namespace rnn_vad {
+namespace {
+
+constexpr int kAutoCorrelationFftOrder = 9; // Length-512 FFT.
+static_assert(1 << kAutoCorrelationFftOrder >
+ kNumInvertedLags12kHz + kBufSize12kHz - kMaxPitch12kHz,
+ "");
+
+} // namespace
+
+AutoCorrelationCalculator::AutoCorrelationCalculator()
+ : fft_(1 << kAutoCorrelationFftOrder, Pffft::FftType::kReal),
+ tmp_(fft_.CreateBuffer()),
+ X_(fft_.CreateBuffer()),
+ H_(fft_.CreateBuffer()) {}
+
+AutoCorrelationCalculator::~AutoCorrelationCalculator() = default;
+
+// The auto-correlations coefficients are computed as follows:
+// |.........|...........| <- pitch buffer
+// [ x (fixed) ]
+// [ y_0 ]
+// [ y_{m-1} ]
+// x and y are sub-array of equal length; x is never moved, whereas y slides.
+// The cross-correlation between y_0 and x corresponds to the auto-correlation
+// for the maximum pitch period. Hence, the first value in |auto_corr| has an
+// inverted lag equal to 0 that corresponds to a lag equal to the maximum
+// pitch period.
+void AutoCorrelationCalculator::ComputeOnPitchBuffer(
+ rtc::ArrayView<const float, kBufSize12kHz> pitch_buf,
+ rtc::ArrayView<float, kNumInvertedLags12kHz> auto_corr) {
+ RTC_DCHECK_LT(auto_corr.size(), kMaxPitch12kHz);
+ RTC_DCHECK_GT(pitch_buf.size(), kMaxPitch12kHz);
+ constexpr size_t kFftFrameSize = 1 << kAutoCorrelationFftOrder;
+ constexpr size_t kConvolutionLength = kBufSize12kHz - kMaxPitch12kHz;
+ static_assert(kConvolutionLength == kFrameSize20ms12kHz,
+ "Mismatch between pitch buffer size, frame size and maximum "
+ "pitch period.");
+ static_assert(kFftFrameSize > kNumInvertedLags12kHz + kConvolutionLength,
+ "The FFT length is not sufficiently big to avoid cyclic "
+ "convolution errors.");
+ auto tmp = tmp_->GetView();
+
+ // Compute the FFT for the reversed reference frame - i.e.,
+ // pitch_buf[-kConvolutionLength:].
+ std::reverse_copy(pitch_buf.end() - kConvolutionLength, pitch_buf.end(),
+ tmp.begin());
+ std::fill(tmp.begin() + kConvolutionLength, tmp.end(), 0.f);
+ fft_.ForwardTransform(*tmp_, H_.get(), /*ordered=*/false);
+
+ // Compute the FFT for the sliding frames chunk. The sliding frames are
+ // defined as pitch_buf[i:i+kConvolutionLength] where i in
+ // [0, kNumInvertedLags12kHz). The chunk includes all of them, hence it is
+ // defined as pitch_buf[:kNumInvertedLags12kHz+kConvolutionLength].
+ std::copy(pitch_buf.begin(),
+ pitch_buf.begin() + kConvolutionLength + kNumInvertedLags12kHz,
+ tmp.begin());
+ std::fill(tmp.begin() + kNumInvertedLags12kHz + kConvolutionLength, tmp.end(),
+ 0.f);
+ fft_.ForwardTransform(*tmp_, X_.get(), /*ordered=*/false);
+
+ // Convolve in the frequency domain.
+ constexpr float kScalingFactor = 1.f / static_cast<float>(kFftFrameSize);
+ std::fill(tmp.begin(), tmp.end(), 0.f);
+ fft_.FrequencyDomainConvolve(*X_, *H_, tmp_.get(), kScalingFactor);
+ fft_.BackwardTransform(*tmp_, tmp_.get(), /*ordered=*/false);
+
+ // Extract the auto-correlation coefficients.
+ std::copy(tmp.begin() + kConvolutionLength - 1,
+ tmp.begin() + kConvolutionLength + kNumInvertedLags12kHz - 1,
+ auto_corr.begin());
+}
+
+} // namespace rnn_vad
+} // namespace webrtc
diff --git a/modules/audio_processing/agc2/rnn_vad/auto_correlation.h b/modules/audio_processing/agc2/rnn_vad/auto_correlation.h
new file mode 100644
index 0000000..de7f453
--- /dev/null
+++ b/modules/audio_processing/agc2/rnn_vad/auto_correlation.h
@@ -0,0 +1,49 @@
+/*
+ * Copyright (c) 2019 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_AGC2_RNN_VAD_AUTO_CORRELATION_H_
+#define MODULES_AUDIO_PROCESSING_AGC2_RNN_VAD_AUTO_CORRELATION_H_
+
+#include <memory>
+
+#include "api/array_view.h"
+#include "modules/audio_processing/agc2/rnn_vad/common.h"
+#include "modules/audio_processing/utility/pffft_wrapper.h"
+
+namespace webrtc {
+namespace rnn_vad {
+
+// Class to compute the auto correlation on the pitch buffer for a target pitch
+// interval.
+class AutoCorrelationCalculator {
+ public:
+ AutoCorrelationCalculator();
+ AutoCorrelationCalculator(const AutoCorrelationCalculator&) = delete;
+ AutoCorrelationCalculator& operator=(const AutoCorrelationCalculator&) =
+ delete;
+ ~AutoCorrelationCalculator();
+
+ // Computes the auto-correlation coefficients for a target pitch interval.
+ // |auto_corr| indexes are inverted lags.
+ void ComputeOnPitchBuffer(
+ rtc::ArrayView<const float, kBufSize12kHz> pitch_buf,
+ rtc::ArrayView<float, kNumInvertedLags12kHz> auto_corr);
+
+ private:
+ Pffft fft_;
+ std::unique_ptr<Pffft::FloatBuffer> tmp_;
+ std::unique_ptr<Pffft::FloatBuffer> X_;
+ std::unique_ptr<Pffft::FloatBuffer> H_;
+};
+
+} // namespace rnn_vad
+} // namespace webrtc
+
+#endif // MODULES_AUDIO_PROCESSING_AGC2_RNN_VAD_AUTO_CORRELATION_H_
diff --git a/modules/audio_processing/agc2/rnn_vad/auto_correlation_unittest.cc b/modules/audio_processing/agc2/rnn_vad/auto_correlation_unittest.cc
new file mode 100644
index 0000000..a5e456a
--- /dev/null
+++ b/modules/audio_processing/agc2/rnn_vad/auto_correlation_unittest.cc
@@ -0,0 +1,62 @@
+/*
+ * Copyright (c) 2019 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/agc2/rnn_vad/auto_correlation.h"
+
+#include "modules/audio_processing/agc2/rnn_vad/pitch_search_internal.h"
+#include "modules/audio_processing/agc2/rnn_vad/test_utils.h"
+#include "test/gtest.h"
+
+namespace webrtc {
+namespace rnn_vad {
+namespace test {
+
+TEST(RnnVadTest, PitchBufferAutoCorrelationWithinTolerance) {
+ PitchTestData test_data;
+ std::array<float, kBufSize12kHz> pitch_buf_decimated;
+ Decimate2x(test_data.GetPitchBufView(), pitch_buf_decimated);
+ std::array<float, kNumPitchBufAutoCorrCoeffs> computed_output;
+ {
+ // TODO(bugs.webrtc.org/8948): Add when the issue is fixed.
+ // FloatingPointExceptionObserver fpe_observer;
+ AutoCorrelationCalculator auto_corr_calculator;
+ auto_corr_calculator.ComputeOnPitchBuffer(pitch_buf_decimated,
+ computed_output);
+ }
+ auto auto_corr_view = test_data.GetPitchBufAutoCorrCoeffsView();
+ ExpectNearAbsolute({auto_corr_view.data(), auto_corr_view.size()},
+ computed_output, 3e-3f);
+}
+
+// Check that the auto correlation function computes the right thing for a
+// simple use case.
+TEST(RnnVadTest, CheckAutoCorrelationOnConstantPitchBuffer) {
+ // Create constant signal with no pitch.
+ std::array<float, kBufSize12kHz> pitch_buf_decimated;
+ std::fill(pitch_buf_decimated.begin(), pitch_buf_decimated.end(), 1.f);
+ std::array<float, kNumPitchBufAutoCorrCoeffs> computed_output;
+ {
+ // TODO(bugs.webrtc.org/8948): Add when the issue is fixed.
+ // FloatingPointExceptionObserver fpe_observer;
+ AutoCorrelationCalculator auto_corr_calculator;
+ auto_corr_calculator.ComputeOnPitchBuffer(pitch_buf_decimated,
+ computed_output);
+ }
+ // The expected output is constantly the length of the fixed 'x'
+ // array in ComputePitchAutoCorrelation.
+ std::array<float, kNumPitchBufAutoCorrCoeffs> expected_output;
+ std::fill(expected_output.begin(), expected_output.end(),
+ kBufSize12kHz - kMaxPitch12kHz);
+ ExpectNearAbsolute(expected_output, computed_output, 4e-5f);
+}
+
+} // namespace test
+} // namespace rnn_vad
+} // namespace webrtc
diff --git a/modules/audio_processing/agc2/rnn_vad/common.h b/modules/audio_processing/agc2/rnn_vad/common.h
index b98438d..2f16cd4 100644
--- a/modules/audio_processing/agc2/rnn_vad/common.h
+++ b/modules/audio_processing/agc2/rnn_vad/common.h
@@ -20,18 +20,21 @@
constexpr size_t kFrameSize10ms24kHz = kSampleRate24kHz / 100;
constexpr size_t kFrameSize20ms24kHz = kFrameSize10ms24kHz * 2;
-// Pitch analysis params.
+// Pitch buffer.
constexpr size_t kMinPitch24kHz = kSampleRate24kHz / 800; // 0.00125 s.
constexpr size_t kMaxPitch24kHz = kSampleRate24kHz / 62.5; // 0.016 s.
constexpr size_t kBufSize24kHz = kMaxPitch24kHz + kFrameSize20ms24kHz;
static_assert((kBufSize24kHz & 1) == 0, "The buffer size must be even.");
+// 24 kHz analysis.
// Define a higher minimum pitch period for the initial search. This is used to
// avoid searching for very short periods, for which a refinement step is
// responsible.
constexpr size_t kInitialMinPitch24kHz = 3 * kMinPitch24kHz;
static_assert(kMinPitch24kHz < kInitialMinPitch24kHz, "");
static_assert(kInitialMinPitch24kHz < kMaxPitch24kHz, "");
+static_assert(kMaxPitch24kHz > kInitialMinPitch24kHz, "");
+constexpr size_t kNumInvertedLags24kHz = kMaxPitch24kHz - kInitialMinPitch24kHz;
// 12 kHz analysis.
constexpr size_t kSampleRate12kHz = 12000;
@@ -40,6 +43,10 @@
constexpr size_t kBufSize12kHz = kBufSize24kHz / 2;
constexpr size_t kInitialMinPitch12kHz = kInitialMinPitch24kHz / 2;
constexpr size_t kMaxPitch12kHz = kMaxPitch24kHz / 2;
+static_assert(kMaxPitch12kHz > kInitialMinPitch12kHz, "");
+// The inverted lags for the pitch interval [|kInitialMinPitch12kHz|,
+// |kMaxPitch12kHz|] are in the range [0, |kNumInvertedLags12kHz|].
+constexpr size_t kNumInvertedLags12kHz = kMaxPitch12kHz - kInitialMinPitch12kHz;
// 48 kHz constants.
constexpr size_t kMinPitch48kHz = kMinPitch24kHz * 2;
diff --git a/modules/audio_processing/agc2/rnn_vad/pitch_search.cc b/modules/audio_processing/agc2/rnn_vad/pitch_search.cc
index aa0b751..1b3b459 100644
--- a/modules/audio_processing/agc2/rnn_vad/pitch_search.cc
+++ b/modules/audio_processing/agc2/rnn_vad/pitch_search.cc
@@ -19,8 +19,7 @@
namespace rnn_vad {
PitchEstimator::PitchEstimator()
- : fft_(RealFourier::Create(kAutoCorrelationFftOrder)),
- pitch_buf_decimated_(kBufSize12kHz),
+ : pitch_buf_decimated_(kBufSize12kHz),
pitch_buf_decimated_view_(pitch_buf_decimated_.data(), kBufSize12kHz),
auto_corr_(kNumInvertedLags12kHz),
auto_corr_view_(auto_corr_.data(), kNumInvertedLags12kHz) {
@@ -34,20 +33,16 @@
rtc::ArrayView<const float, kBufSize24kHz> pitch_buf) {
// Perform the initial pitch search at 12 kHz.
Decimate2x(pitch_buf, pitch_buf_decimated_view_);
- // Compute auto-correlation terms.
- ComputePitchAutoCorrelation(pitch_buf_decimated_view_, kMaxPitch12kHz,
- auto_corr_view_, fft_.get());
-
- // Search for pitch at 12 kHz.
+ auto_corr_calculator_.ComputeOnPitchBuffer(pitch_buf_decimated_view_,
+ auto_corr_view_);
std::array<size_t, 2> pitch_candidates_inv_lags = FindBestPitchPeriods(
auto_corr_view_, pitch_buf_decimated_view_, kMaxPitch12kHz);
-
// Refine the pitch period estimation.
// The refinement is done using the pitch buffer that contains 24 kHz samples.
// Therefore, adapt the inverted lags in |pitch_candidates_inv_lags| from 12
// to 24 kHz.
- for (size_t i = 0; i < pitch_candidates_inv_lags.size(); ++i)
- pitch_candidates_inv_lags[i] *= 2;
+ pitch_candidates_inv_lags[0] *= 2;
+ pitch_candidates_inv_lags[1] *= 2;
size_t pitch_inv_lag_48kHz =
RefinePitchPeriod48kHz(pitch_buf, pitch_candidates_inv_lags);
// Look for stronger harmonics to find the final pitch period and its gain.
diff --git a/modules/audio_processing/agc2/rnn_vad/pitch_search.h b/modules/audio_processing/agc2/rnn_vad/pitch_search.h
index 5914535..74133d0 100644
--- a/modules/audio_processing/agc2/rnn_vad/pitch_search.h
+++ b/modules/audio_processing/agc2/rnn_vad/pitch_search.h
@@ -15,7 +15,7 @@
#include <vector>
#include "api/array_view.h"
-#include "common_audio/real_fourier.h"
+#include "modules/audio_processing/agc2/rnn_vad/auto_correlation.h"
#include "modules/audio_processing/agc2/rnn_vad/common.h"
#include "modules/audio_processing/agc2/rnn_vad/pitch_info.h"
#include "modules/audio_processing/agc2/rnn_vad/pitch_search_internal.h"
@@ -36,7 +36,7 @@
private:
PitchInfo last_pitch_48kHz_;
- std::unique_ptr<RealFourier> fft_;
+ AutoCorrelationCalculator auto_corr_calculator_;
std::vector<float> pitch_buf_decimated_;
rtc::ArrayView<float, kBufSize12kHz> pitch_buf_decimated_view_;
std::vector<float> auto_corr_;
diff --git a/modules/audio_processing/agc2/rnn_vad/pitch_search_internal.cc b/modules/audio_processing/agc2/rnn_vad/pitch_search_internal.cc
index 7c17dfb..0561c37 100644
--- a/modules/audio_processing/agc2/rnn_vad/pitch_search_internal.cc
+++ b/modules/audio_processing/agc2/rnn_vad/pitch_search_internal.cc
@@ -13,7 +13,6 @@
#include <stdlib.h>
#include <algorithm>
#include <cmath>
-#include <complex>
#include <cstddef>
#include <numeric>
@@ -213,64 +212,6 @@
}
}
-void ComputePitchAutoCorrelation(
- rtc::ArrayView<const float, kBufSize12kHz> pitch_buf,
- size_t max_pitch_period,
- rtc::ArrayView<float, kNumInvertedLags12kHz> auto_corr,
- webrtc::RealFourier* fft) {
- RTC_DCHECK_GT(max_pitch_period, auto_corr.size());
- RTC_DCHECK_LT(max_pitch_period, pitch_buf.size());
- RTC_DCHECK(fft);
-
- constexpr size_t time_domain_fft_length = 1 << kAutoCorrelationFftOrder;
- constexpr size_t freq_domain_fft_length = time_domain_fft_length / 2 + 1;
-
- RTC_DCHECK_EQ(RealFourier::FftLength(fft->order()), time_domain_fft_length);
- RTC_DCHECK_EQ(RealFourier::ComplexLength(fft->order()),
- freq_domain_fft_length);
-
- // Cross-correlation of y_i=pitch_buf[i:i+convolution_length] and
- // x=pitch_buf[-convolution_length:] is equivalent to convolution of
- // y_i and reversed(x). New notation: h=reversed(x), x=y.
- std::array<float, time_domain_fft_length> h{};
- std::array<float, time_domain_fft_length> x{};
-
- const size_t convolution_length = kBufSize12kHz - max_pitch_period;
- // Check that the FFT-length is big enough to avoid cyclic
- // convolution errors.
- RTC_DCHECK_GT(time_domain_fft_length,
- kNumInvertedLags12kHz + convolution_length);
-
- // h[0:convolution_length] is reversed pitch_buf[-convolution_length:].
- std::reverse_copy(pitch_buf.end() - convolution_length, pitch_buf.end(),
- h.begin());
-
- // x is pitch_buf[:kNumInvertedLags12kHz + convolution_length].
- std::copy(pitch_buf.begin(),
- pitch_buf.begin() + kNumInvertedLags12kHz + convolution_length,
- x.begin());
-
- // Shift to frequency domain.
- std::array<std::complex<float>, freq_domain_fft_length> X{};
- std::array<std::complex<float>, freq_domain_fft_length> H{};
- fft->Forward(&x[0], &X[0]);
- fft->Forward(&h[0], &H[0]);
-
- // Convolve in frequency domain.
- for (size_t i = 0; i < X.size(); ++i) {
- X[i] *= H[i];
- }
-
- // Shift back to time domain.
- std::array<float, time_domain_fft_length> x_conv_h;
- fft->Inverse(&X[0], &x_conv_h[0]);
-
- // Collect the result.
- std::copy(x_conv_h.begin() + convolution_length - 1,
- x_conv_h.begin() + convolution_length + kNumInvertedLags12kHz - 1,
- auto_corr.begin());
-}
-
std::array<size_t, 2> FindBestPitchPeriods(
rtc::ArrayView<const float> auto_corr,
rtc::ArrayView<const float> pitch_buf,
diff --git a/modules/audio_processing/agc2/rnn_vad/pitch_search_internal.h b/modules/audio_processing/agc2/rnn_vad/pitch_search_internal.h
index aabf713..6ccd165 100644
--- a/modules/audio_processing/agc2/rnn_vad/pitch_search_internal.h
+++ b/modules/audio_processing/agc2/rnn_vad/pitch_search_internal.h
@@ -15,25 +15,12 @@
#include <array>
#include "api/array_view.h"
-#include "common_audio/real_fourier.h"
#include "modules/audio_processing/agc2/rnn_vad/common.h"
#include "modules/audio_processing/agc2/rnn_vad/pitch_info.h"
namespace webrtc {
namespace rnn_vad {
-// The inverted lags for the pitch interval [|kInitialMinPitch12kHz|,
-// |kMaxPitch12kHz|] are in the range [0, |kNumInvertedLags|].
-static_assert(kMaxPitch12kHz > kInitialMinPitch12kHz, "");
-static_assert(kMaxPitch24kHz > kInitialMinPitch24kHz, "");
-constexpr size_t kNumInvertedLags12kHz = kMaxPitch12kHz - kInitialMinPitch12kHz;
-constexpr size_t kNumInvertedLags24kHz = kMaxPitch24kHz - kInitialMinPitch24kHz;
-constexpr int kAutoCorrelationFftOrder = 9; // Length-512 FFT.
-
-static_assert(1 << kAutoCorrelationFftOrder >
- kNumInvertedLags12kHz + kBufSize12kHz - kMaxPitch12kHz,
- "");
-
// Performs 2x decimation without any anti-aliasing filter.
void Decimate2x(rtc::ArrayView<const float, kBufSize24kHz> src,
rtc::ArrayView<float, kBufSize12kHz> dst);
@@ -61,25 +48,6 @@
rtc::ArrayView<const float, kBufSize24kHz> pitch_buf,
rtc::ArrayView<float, kMaxPitch24kHz + 1> yy_values);
-// Computes the auto-correlation coefficients for a given pitch interval.
-// |auto_corr| indexes are inverted lags.
-//
-// The auto-correlations coefficients are computed as follows:
-// |.........|...........| <- pitch buffer
-// [ x (fixed) ]
-// [ y_0 ]
-// [ y_{m-1} ]
-// x and y are sub-array of equal length; x is never moved, whereas y slides.
-// The cross-correlation between y_0 and x corresponds to the auto-correlation
-// for the maximum pitch period. Hence, the first value in |auto_corr| has an
-// inverted lag equal to 0 that corresponds to a lag equal to the maximum pitch
-// period.
-void ComputePitchAutoCorrelation(
- rtc::ArrayView<const float, kBufSize12kHz> pitch_buf,
- size_t max_pitch_period,
- rtc::ArrayView<float, kNumInvertedLags12kHz> auto_corr,
- webrtc::RealFourier* fft);
-
// Given the auto-correlation coefficients stored according to
// ComputePitchAutoCorrelation() (i.e., using inverted lags), returns the best
// and the second best pitch periods.
diff --git a/modules/audio_processing/agc2/rnn_vad/pitch_search_internal_unittest.cc b/modules/audio_processing/agc2/rnn_vad/pitch_search_internal_unittest.cc
index 8ff6ac1..bd2ea24 100644
--- a/modules/audio_processing/agc2/rnn_vad/pitch_search_internal_unittest.cc
+++ b/modules/audio_processing/agc2/rnn_vad/pitch_search_internal_unittest.cc
@@ -9,7 +9,6 @@
*/
#include "modules/audio_processing/agc2/rnn_vad/pitch_search_internal.h"
-#include "common_audio/real_fourier.h"
#include <array>
#include <tuple>
@@ -30,34 +29,6 @@
};
constexpr std::array<float, 2> kTestPitchGains = {0.35f, 0.75f};
-constexpr size_t kNumPitchBufSquareEnergies = 385;
-constexpr size_t kNumPitchBufAutoCorrCoeffs = 147;
-constexpr size_t kTestDataSize =
- kBufSize24kHz + kNumPitchBufSquareEnergies + kNumPitchBufAutoCorrCoeffs;
-
-class TestData {
- public:
- TestData() {
- auto test_data_reader = CreatePitchSearchTestDataReader();
- test_data_reader->ReadChunk(test_data_);
- }
- rtc::ArrayView<const float, kBufSize24kHz> GetPitchBufView() {
- return {test_data_.data(), kBufSize24kHz};
- }
- rtc::ArrayView<const float, kNumPitchBufSquareEnergies>
- GetPitchBufSquareEnergiesView() {
- return {test_data_.data() + kBufSize24kHz, kNumPitchBufSquareEnergies};
- }
- rtc::ArrayView<const float, kNumPitchBufAutoCorrCoeffs>
- GetPitchBufAutoCorrCoeffsView() {
- return {test_data_.data() + kBufSize24kHz + kNumPitchBufSquareEnergies,
- kNumPitchBufAutoCorrCoeffs};
- }
-
- private:
- std::array<float, kTestDataSize> test_data_;
-};
-
} // namespace
class ComputePitchGainThresholdTest
@@ -107,7 +78,7 @@
std::make_tuple(78, 2, 156, 0.72750503f, 153, 0.85069299f, 0.618379f)));
TEST(RnnVadTest, ComputeSlidingFrameSquareEnergiesBitExactness) {
- TestData test_data;
+ PitchTestData test_data;
std::array<float, kNumPitchBufSquareEnergies> computed_output;
{
// TODO(bugs.webrtc.org/8948): Add when the issue is fixed.
@@ -120,51 +91,8 @@
computed_output, 3e-2f);
}
-TEST(RnnVadTest, ComputePitchAutoCorrelationBitExactness) {
- TestData test_data;
- std::array<float, kBufSize12kHz> pitch_buf_decimated;
- Decimate2x(test_data.GetPitchBufView(), pitch_buf_decimated);
- std::array<float, kNumPitchBufAutoCorrCoeffs> computed_output;
- {
- // TODO(bugs.webrtc.org/8948): Add when the issue is fixed.
- // FloatingPointExceptionObserver fpe_observer;
- std::unique_ptr<RealFourier> fft =
- RealFourier::Create(kAutoCorrelationFftOrder);
- ComputePitchAutoCorrelation(pitch_buf_decimated, kMaxPitch12kHz,
- computed_output, fft.get());
- }
- auto auto_corr_view = test_data.GetPitchBufAutoCorrCoeffsView();
- ExpectNearAbsolute({auto_corr_view.data(), auto_corr_view.size()},
- computed_output, 3e-3f);
-}
-
-// Check that the auto correlation function computes the right thing for a
-// simple use case.
-TEST(RnnVadTest, ComputePitchAutoCorrelationConstantBuffer) {
- // Create constant signal with no pitch.
- std::array<float, kBufSize12kHz> pitch_buf_decimated;
- std::fill(pitch_buf_decimated.begin(), pitch_buf_decimated.end(), 1.f);
-
- std::array<float, kNumPitchBufAutoCorrCoeffs> computed_output;
- {
- // TODO(bugs.webrtc.org/8948): Add when the issue is fixed.
- // FloatingPointExceptionObserver fpe_observer;
- std::unique_ptr<RealFourier> fft =
- RealFourier::Create(kAutoCorrelationFftOrder);
- ComputePitchAutoCorrelation(pitch_buf_decimated, kMaxPitch12kHz,
- computed_output, fft.get());
- }
-
- // The expected output is constantly the length of the fixed 'x'
- // array in ComputePitchAutoCorrelation.
- std::array<float, kNumPitchBufAutoCorrCoeffs> expected_output;
- std::fill(expected_output.begin(), expected_output.end(),
- kBufSize12kHz - kMaxPitch12kHz);
- ExpectNearAbsolute(expected_output, computed_output, 4e-5f);
-}
-
TEST(RnnVadTest, FindBestPitchPeriodsBitExactness) {
- TestData test_data;
+ PitchTestData test_data;
std::array<float, kBufSize12kHz> pitch_buf_decimated;
Decimate2x(test_data.GetPitchBufView(), pitch_buf_decimated);
std::array<size_t, 2> pitch_candidates_inv_lags;
@@ -181,7 +109,7 @@
}
TEST(RnnVadTest, RefinePitchPeriod48kHzBitExactness) {
- TestData test_data;
+ PitchTestData test_data;
std::array<float, kBufSize12kHz> pitch_buf_decimated;
Decimate2x(test_data.GetPitchBufView(), pitch_buf_decimated);
size_t pitch_inv_lag;
@@ -207,7 +135,7 @@
const float prev_pitch_gain = std::get<2>(params);
const int expected_pitch_period = std::get<3>(params);
const float expected_pitch_gain = std::get<4>(params);
- TestData test_data;
+ PitchTestData test_data;
{
// TODO(bugs.webrtc.org/8948): Add when the issue is fixed.
// FloatingPointExceptionObserver fpe_observer;
diff --git a/modules/audio_processing/agc2/rnn_vad/test_utils.cc b/modules/audio_processing/agc2/rnn_vad/test_utils.cc
index 8decbd0..4dae8cd 100644
--- a/modules/audio_processing/agc2/rnn_vad/test_utils.cc
+++ b/modules/audio_processing/agc2/rnn_vad/test_utils.cc
@@ -111,6 +111,28 @@
return {std::move(ptr), ptr->data_length()};
}
+PitchTestData::PitchTestData() {
+ auto test_data_reader = CreatePitchSearchTestDataReader();
+ test_data_reader->ReadChunk(test_data_);
+}
+
+PitchTestData::~PitchTestData() = default;
+
+rtc::ArrayView<const float, kBufSize24kHz> PitchTestData::GetPitchBufView() {
+ return {test_data_.data(), kBufSize24kHz};
+}
+
+rtc::ArrayView<const float, kNumPitchBufSquareEnergies>
+PitchTestData::GetPitchBufSquareEnergiesView() {
+ return {test_data_.data() + kBufSize24kHz, kNumPitchBufSquareEnergies};
+}
+
+rtc::ArrayView<const float, kNumPitchBufAutoCorrCoeffs>
+PitchTestData::GetPitchBufAutoCorrCoeffsView() {
+ return {test_data_.data() + kBufSize24kHz + kNumPitchBufSquareEnergies,
+ kNumPitchBufAutoCorrCoeffs};
+}
+
} // namespace test
} // namespace rnn_vad
} // namespace webrtc
diff --git a/modules/audio_processing/agc2/rnn_vad/test_utils.h b/modules/audio_processing/agc2/rnn_vad/test_utils.h
index 15be85a..f9d7376 100644
--- a/modules/audio_processing/agc2/rnn_vad/test_utils.h
+++ b/modules/audio_processing/agc2/rnn_vad/test_utils.h
@@ -12,6 +12,7 @@
#define MODULES_AUDIO_PROCESSING_AGC2_RNN_VAD_TEST_UTILS_H_
#include <algorithm>
+#include <array>
#include <fstream>
#include <limits>
#include <memory>
@@ -20,6 +21,7 @@
#include <vector>
#include "api/array_view.h"
+#include "modules/audio_processing/agc2/rnn_vad/common.h"
#include "rtc_base/checks.h"
namespace webrtc {
@@ -118,6 +120,27 @@
std::pair<std::unique_ptr<BinaryFileReader<float>>, const size_t>
CreateVadProbsReader();
+constexpr size_t kNumPitchBufAutoCorrCoeffs = 147;
+constexpr size_t kNumPitchBufSquareEnergies = 385;
+constexpr size_t kPitchTestDataSize =
+ kBufSize24kHz + kNumPitchBufSquareEnergies + kNumPitchBufAutoCorrCoeffs;
+
+// Class to retrieve a test pitch buffer content and the expected output for the
+// analysis steps.
+class PitchTestData {
+ public:
+ PitchTestData();
+ ~PitchTestData();
+ rtc::ArrayView<const float, kBufSize24kHz> GetPitchBufView();
+ rtc::ArrayView<const float, kNumPitchBufSquareEnergies>
+ GetPitchBufSquareEnergiesView();
+ rtc::ArrayView<const float, kNumPitchBufAutoCorrCoeffs>
+ GetPitchBufAutoCorrCoeffsView();
+
+ private:
+ std::array<float, kPitchTestDataSize> test_data_;
+};
+
} // namespace test
} // namespace rnn_vad
} // namespace webrtc
