pitch_search_internal.cc | Explore in Territory

/*
 *  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/agc2/rnn_vad/pitch_search_internal.h"

#include <stdlib.h>

#include <algorithm>
#include <cmath>
#include <cstddef>
#include <numeric>

#include "modules/audio_processing/agc2/rnn_vad/common.h"
#include "modules/audio_processing/agc2/rnn_vad/vector_math.h"
#include "rtc_base/checks.h"
#include "rtc_base/numerics/safe_compare.h"
#include "rtc_base/numerics/safe_conversions.h"
#include "rtc_base/system/arch.h"

namespace webrtc {
namespace rnn_vad {
namespace {

float ComputeAutoCorrelation(
    int inverted_lag,
    rtc::ArrayView<const float, kBufSize24kHz> pitch_buffer,
    const VectorMath& vector_math) { … }

// Given an auto-correlation coefficient `curr_auto_correlation` and its
// neighboring values `prev_auto_correlation` and `next_auto_correlation`
// computes a pseudo-interpolation offset to be applied to the pitch period
// associated to `curr`. The output is a lag in {-1, 0, +1}.
// TODO(bugs.webrtc.org/9076): Consider removing this method.
// `GetPitchPseudoInterpolationOffset()` it is relevant only if the spectral
// analysis works at a sample rate that is twice as that of the pitch buffer;
// In particular, it is not relevant for the estimated pitch period feature fed
// into the RNN.
int GetPitchPseudoInterpolationOffset(float prev_auto_correlation,
                                      float curr_auto_correlation,
                                      float next_auto_correlation) { … }

// Refines a pitch period `lag` encoded as lag with pseudo-interpolation. The
// output sample rate is twice as that of `lag`.
int PitchPseudoInterpolationLagPitchBuf(
    int lag,
    rtc::ArrayView<const float, kBufSize24kHz> pitch_buffer,
    const VectorMath& vector_math) { … }

// Integer multipliers used in ComputeExtendedPitchPeriod48kHz() when
// looking for sub-harmonics.
// The values have been chosen to serve the following algorithm. Given the
// initial pitch period T, we examine whether one of its harmonics is the true
// fundamental frequency. We consider T/k with k in {2, ..., 15}. For each of
// these harmonics, in addition to the pitch strength of itself, we choose one
// multiple of its pitch period, n*T/k, to validate it (by averaging their pitch
// strengths). The multiplier n is chosen so that n*T/k is used only one time
// over all k. When for example k = 4, we should also expect a peak at 3*T/4.
// When k = 8 instead we don't want to look at 2*T/8, since we have already
// checked T/4 before. Instead, we look at T*3/8.
// The array can be generate in Python as follows:
//   from fractions import Fraction
//   # Smallest positive integer not in X.
//   def mex(X):
//     for i in range(1, int(max(X)+2)):
//       if i not in X:
//         return i
//   # Visited multiples of the period.
//   S = {1}
//   for n in range(2, 16):
//     sn = mex({n * i for i in S} | {1})
//     S = S | {Fraction(1, n), Fraction(sn, n)}
//     print(sn, end=', ')
constexpr std::array<int, 14> kSubHarmonicMultipliers = …;

struct Range { … };

// Number of analyzed pitches to the left(right) of a pitch candidate.
constexpr int kPitchNeighborhoodRadius = …;

// Creates a pitch period interval centered in `inverted_lag` with hard-coded
// radius. Clipping is applied so that the interval is always valid for a 24 kHz
// pitch buffer.
Range CreateInvertedLagRange(int inverted_lag) { … }

constexpr int kNumPitchCandidates = …;  // Best and second best.
// Maximum number of analyzed pitch periods.
constexpr int kMaxPitchPeriods24kHz = …;

// Collection of inverted lags.
class InvertedLagsIndex { … };

// Computes the auto correlation coefficients for the inverted lags in the
// closed interval `inverted_lags`. Updates `inverted_lags_index` by appending
// the inverted lags for the computed auto correlation values.
void ComputeAutoCorrelation(
    Range inverted_lags,
    rtc::ArrayView<const float, kBufSize24kHz> pitch_buffer,
    rtc::ArrayView<float, kInitialNumLags24kHz> auto_correlation,
    InvertedLagsIndex& inverted_lags_index,
    const VectorMath& vector_math) { … }

// Searches the strongest pitch period at 24 kHz and returns its inverted lag at
// 48 kHz.
int ComputePitchPeriod48kHz(
    rtc::ArrayView<const float, kBufSize24kHz> pitch_buffer,
    rtc::ArrayView<const int> inverted_lags,
    rtc::ArrayView<const float, kInitialNumLags24kHz> auto_correlation,
    rtc::ArrayView<const float, kRefineNumLags24kHz> y_energy,
    const VectorMath& vector_math) { … }

// Returns an alternative pitch period for `pitch_period` given a `multiplier`
// and a `divisor` of the period.
constexpr int GetAlternativePitchPeriod(int pitch_period,
                                        int multiplier,
                                        int divisor) { … }

// Returns true if the alternative pitch period is stronger than the initial one
// given the last estimated pitch and the value of `period_divisor` used to
// compute the alternative pitch period via `GetAlternativePitchPeriod()`.
bool IsAlternativePitchStrongerThanInitial(PitchInfo last,
                                           PitchInfo initial,
                                           PitchInfo alternative,
                                           int period_divisor) { … }

}  // namespace

void Decimate2x(rtc::ArrayView<const float, kBufSize24kHz> src,
                rtc::ArrayView<float, kBufSize12kHz> dst) { … }

void ComputeSlidingFrameSquareEnergies24kHz(
    rtc::ArrayView<const float, kBufSize24kHz> pitch_buffer,
    rtc::ArrayView<float, kRefineNumLags24kHz> y_energy,
    AvailableCpuFeatures cpu_features) { … }

CandidatePitchPeriods ComputePitchPeriod12kHz(
    rtc::ArrayView<const float, kBufSize12kHz> pitch_buffer,
    rtc::ArrayView<const float, kNumLags12kHz> auto_correlation,
    AvailableCpuFeatures cpu_features) { … }

int ComputePitchPeriod48kHz(
    rtc::ArrayView<const float, kBufSize24kHz> pitch_buffer,
    rtc::ArrayView<const float, kRefineNumLags24kHz> y_energy,
    CandidatePitchPeriods pitch_candidates,
    AvailableCpuFeatures cpu_features) { … }

PitchInfo ComputeExtendedPitchPeriod48kHz(
    rtc::ArrayView<const float, kBufSize24kHz> pitch_buffer,
    rtc::ArrayView<const float, kRefineNumLags24kHz> y_energy,
    int initial_pitch_period_48kHz,
    PitchInfo last_pitch_48kHz,
    AvailableCpuFeatures cpu_features) { … }

}  // namespace rnn_vad
}  // namespace webrtc
chromium/third_party/webrtc/modules/audio_processing/agc2/rnn_vad/pitch_search_internal.cc
