blob: 3148f1b3ff2e58cbdea42f8f0498813d2111e5e4 [file] [log] [blame]
* 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 <stddef.h>
#include <sys/types.h>
#include <array>
#include <vector>
#include "api/array_view.h"
#include "modules/audio_processing/agc2/cpu_features.h"
#include "modules/audio_processing/agc2/rnn_vad/common.h"
#include "modules/audio_processing/agc2/rnn_vad/rnn_fc.h"
#include "modules/audio_processing/agc2/rnn_vad/rnn_gru.h"
namespace webrtc {
namespace rnn_vad {
// Recurrent network with hard-coded architecture and weights for voice activity
// detection.
class RnnVad {
explicit RnnVad(const AvailableCpuFeatures& cpu_features);
RnnVad(const RnnVad&) = delete;
RnnVad& operator=(const RnnVad&) = delete;
void Reset();
// Observes `feature_vector` and `is_silence`, updates the RNN and returns the
// current voice probability.
float ComputeVadProbability(
rtc::ArrayView<const float, kFeatureVectorSize> feature_vector,
bool is_silence);
FullyConnectedLayer input_;
GatedRecurrentLayer hidden_;
FullyConnectedLayer output_;
} // namespace rnn_vad
} // namespace webrtc