投稿
  1. 首页
  2. 技术文章
  3. WebRTC

webrtc 波束模块编译过程

RTC观主 • 来源：音频探险记 • • WebRTC

本文主要介绍webrtc中的波束模块的编译过程，关于波束算法的技术原理将会在下篇文章中介绍。

webrtc是一个极其庞大的项目，里面的文件包含特别复杂。正是因为如此，对还是小白的我造成了极大的困难。刚开始打算采取本方法，将波束模块里的nonlinear_beamformer_test.cc当做主文件，然后编译这个，差啥文件就往vs工程里面补，这样搞了一段时间后，便发现这种方式实在是太过愚蠢。因此有可能为了编译一个波束模块，把整个webrtc的代码都弄过来。后来接触到了cmake，才发现了这个工具的神奇。在webrtcmodulesaudio_processingbeamformer目录下找到了波束模块的关键代码，其中nonlinear_beamformer_test.cc就是波束模块的测试代码。

图片

里面的代码如下：

/*
* Copyright (c) 2014 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 <vector>

#include "gflags/gflags.h"
#include "webrtc/base/checks.h"
#include "webrtc/base/format_macros.h"
#include "webrtc/common_audio/channel_buffer.h"
#include "webrtc/common_audio/wav_file.h"
#include "webrtc/modules/audio_processing/beamformer/nonlinear_beamformer.h"
#include "webrtc/modules/audio_processing/test/test_utils.h"

DEFINE_string(i, "", "The name of the input file to read from.");
DEFINE_string(o, "out.wav", "Name of the output file to write to.");
DEFINE_string(mic_positions, "",
   "Space delimited cartesian coordinates of microphones in meters. "
   "The coordinates of each point are contiguous. "
   "For a two element array: "x1 y1 z1 x2 y2 z2"");

namespace webrtc {
namespace {

const int kChunksPerSecond = 100;
const int kChunkSizeMs = 1000 / kChunksPerSecond;

const char kUsage[] =
   "Command-line tool to run beamforming on WAV files. The signal is passedn"
   "in as a single band, unlike the audio processing interface which splitsn"
   "signals into multiple bands.";

}  // namespace

int main(int argc, char* argv[]) {
 google::SetUsageMessage(kUsage);
 google::ParseCommandLineFlags(&argc, &argv, true);

 WavReader in_file(FLAGS_i);
 WavWriter out_file(FLAGS_o, in_file.sample_rate(), 1);

 const size_t num_mics = in_file.num_channels();
 const std::vector<Point> array_geometry =
     ParseArrayGeometry(FLAGS_mic_positions, num_mics);
 RTC_CHECK_EQ(array_geometry.size(), num_mics);

 NonlinearBeamformer bf(array_geometry);
 bf.Initialize(kChunkSizeMs, in_file.sample_rate());

 printf("Input file: %snChannels: %" PRIuS ", Sample rate: %d Hznn",
        FLAGS_i.c_str(), in_file.num_channels(), in_file.sample_rate());
 printf("Output file: %snChannels: %" PRIuS ", Sample rate: %d Hznn",
        FLAGS_o.c_str(), out_file.num_channels(), out_file.sample_rate());

 ChannelBuffer<float> in_buf(
     rtc::CheckedDivExact(in_file.sample_rate(), kChunksPerSecond),
     in_file.num_channels());
 ChannelBuffer<float> out_buf(
     rtc::CheckedDivExact(out_file.sample_rate(), kChunksPerSecond),
     out_file.num_channels());

 std::vector<float> interleaved(in_buf.size());
 while (in_file.ReadSamples(interleaved.size(),
                            &interleaved[0]) == interleaved.size()) {
   FloatS16ToFloat(&interleaved[0], interleaved.size(), &interleaved[0]);
   Deinterleave(&interleaved[0], in_buf.num_frames(),
                in_buf.num_channels(), in_buf.channels());

   bf.ProcessChunk(in_buf, &out_buf);

   Interleave(out_buf.channels(), out_buf.num_frames(),
              out_buf.num_channels(), &interleaved[0]);
   FloatToFloatS16(&interleaved[0], interleaved.size(), &interleaved[0]);
   out_file.WriteSamples(&interleaved[0], interleaved.size());
}

 return 0;
}

}  // namespace webrtc

int main(int argc, char* argv[]) {
 return webrtc::main(argc, argv);
}
根据这个测试文件，大概所需要的关键部分有：解析控制台输入的gflags；读写wav文件的WavReader,WavWriter这两个类都在webrtccommon_audio目录下的wav_file.h中；以及波束类NonlinearBeamformer,它的相关实现在nonlinear_beamformer.cc以及nonlinear_beamformer.h中，其实最核心的代码就是:

bf.ProcessChunk(in_buf, &out_buf)
要想弄明白webrtc的波束算法原理，那么就得弄明白该函数的执行流程以及实现原理。接下来就开始编译波束模块代码了。

首先给出CMakeLists.txt文件.这里面的目录结构是我的机器上的，若在你的机器上编译，可以需要改变相应的目录结构

cmake_minimum_required(VERSION 2.8)

project(wav-beamforming)
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
set(CMAKE_INSTALL_RPATH_USE_LINK_PATH TRUE)

add_subdirectory(gflags)

add_definitions("-DWEBRTC_LINUX -DWEBRTC_POSIX -DWEBRTC_NS_FLOAT")
#-DWEBRTC_UNTRUSTED_DELAY

include_directories(
  webrtc
  webrtc/webrtc/common_audio/signal_processing/include
  webrtc/webrtc/modules/audio_coding/codecs/isac/main/include
  webrtc/webrtc/modules/audio_processing/test/

)

set(WEBRTC_SRC_
  base/buffer.cc
  base/checks.cc
  base/criticalsection.cc
  base/event.cc
  base/event_tracer.cc
  base/logging.cc
  base/platform_file.cc
  base/platform_thread.cc
  base/stringencode.cc
  base/thread_checker_impl.cc
  base/timeutils.cc
  common_audio/audio_converter.cc
  common_audio/audio_ring_buffer.cc
  common_audio/audio_util.cc
  common_audio/blocker.cc
  common_audio/channel_buffer.cc
  common_audio/fft4g.c
  common_audio/fir_filter.cc
  common_audio/fir_filter_sse.cc
  common_audio/lapped_transform.cc
  common_audio/real_fourier.cc
  common_audio/real_fourier_ooura.cc
  common_audio/resampler/push_resampler.cc
  common_audio/resampler/push_sinc_resampler.cc
  common_audio/resampler/resampler.cc
  common_audio/resampler/sinc_resampler.cc
  common_audio/resampler/sinc_resampler_sse.cc
  common_audio/resampler/sinusoidal_linear_chirp_source.cc
  common_audio/ring_buffer.c
  common_audio/signal_processing/auto_correlation.c
  common_audio/signal_processing/auto_corr_to_refl_coef.c
  common_audio/signal_processing/complex_bit_reverse.c
  common_audio/signal_processing/complex_fft.c
  common_audio/signal_processing/copy_set_operations.c
  common_audio/signal_processing/cross_correlation.c
  common_audio/signal_processing/division_operations.c
  common_audio/signal_processing/dot_product_with_scale.c
  common_audio/signal_processing/downsample_fast.c
  common_audio/signal_processing/energy.c
  common_audio/signal_processing/filter_ar.c
  common_audio/signal_processing/filter_ar_fast_q12.c
  common_audio/signal_processing/filter_ma_fast_q12.c
  common_audio/signal_processing/get_hanning_window.c
  common_audio/signal_processing/get_scaling_square.c
  common_audio/signal_processing/ilbc_specific_functions.c
  common_audio/signal_processing/levinson_durbin.c
  common_audio/signal_processing/lpc_to_refl_coef.c
  common_audio/signal_processing/min_max_operations.c
  common_audio/signal_processing/randomization_functions.c
  common_audio/signal_processing/real_fft.c
  common_audio/signal_processing/refl_coef_to_lpc.c
  common_audio/signal_processing/resample_48khz.c
  common_audio/signal_processing/resample_by_2.c
  common_audio/signal_processing/resample_by_2_internal.c
  common_audio/signal_processing/resample.c
  common_audio/signal_processing/resample_fractional.c
  common_audio/signal_processing/spl_init.c
  common_audio/signal_processing/splitting_filter.c
  common_audio/signal_processing/spl_sqrt.c
  common_audio/signal_processing/spl_sqrt_floor.c
  common_audio/signal_processing/sqrt_of_one_minus_x_squared.c
  common_audio/signal_processing/vector_scaling_operations.c
  common_audio/sparse_fir_filter.cc
  common_audio/vad/vad_core.c
  common_audio/vad/vad_filterbank.c
  common_audio/vad/vad_gmm.c
  common_audio/vad/vad_sp.c
  common_audio/vad/webrtc_vad.c
  common_audio/wav_file.cc
  common_audio/wav_header.cc
  common_audio/window_generator.cc
  common_types.cc
  modules/audio_coding/codecs/audio_decoder.cc
  modules/audio_coding/codecs/audio_encoder.cc
  modules/audio_coding/codecs/isac/locked_bandwidth_info.cc
  modules/audio_coding/codecs/isac/main/source/arith_routines.c
  modules/audio_coding/codecs/isac/main/source/arith_routines_hist.c
  modules/audio_coding/codecs/isac/main/source/arith_routines_logist.c
  modules/audio_coding/codecs/isac/main/source/audio_decoder_isac.cc
  modules/audio_coding/codecs/isac/main/source/audio_encoder_isac.cc
  modules/audio_coding/codecs/isac/main/source/bandwidth_estimator.c
  modules/audio_coding/codecs/isac/main/source/crc.c
  modules/audio_coding/codecs/isac/main/source/decode_bwe.c
  modules/audio_coding/codecs/isac/main/source/decode.c
  modules/audio_coding/codecs/isac/main/source/encode.c
  modules/audio_coding/codecs/isac/main/source/encode_lpc_swb.c
  modules/audio_coding/codecs/isac/main/source/entropy_coding.c
  modules/audio_coding/codecs/isac/main/source/fft.c
  modules/audio_coding/codecs/isac/main/source/filterbanks.c
  modules/audio_coding/codecs/isac/main/source/filterbank_tables.c
  modules/audio_coding/codecs/isac/main/source/filter_functions.c
  modules/audio_coding/codecs/isac/main/source/intialize.c
  modules/audio_coding/codecs/isac/main/source/isac.c
  modules/audio_coding/codecs/isac/main/source/lattice.c
  modules/audio_coding/codecs/isac/main/source/lpc_analysis.c
  modules/audio_coding/codecs/isac/main/source/lpc_gain_swb_tables.c
  modules/audio_coding/codecs/isac/main/source/lpc_shape_swb12_tables.c
  modules/audio_coding/codecs/isac/main/source/lpc_shape_swb16_tables.c
  modules/audio_coding/codecs/isac/main/source/lpc_tables.c
  modules/audio_coding/codecs/isac/main/source/pitch_estimator.c
  modules/audio_coding/codecs/isac/main/source/pitch_filter.c
  modules/audio_coding/codecs/isac/main/source/pitch_gain_tables.c
  modules/audio_coding/codecs/isac/main/source/pitch_lag_tables.c
  modules/audio_coding/codecs/isac/main/source/spectrum_ar_model_tables.c
  modules/audio_coding/codecs/isac/main/source/transform.c
  modules/audio_processing/aec/aec_core.cc
  modules/audio_processing/aec/aec_core_sse2.cc
  modules/audio_processing/aec/aec_rdft.cc
  modules/audio_processing/aec/aec_rdft_sse2.cc
  modules/audio_processing/aec/aec_resampler.cc
  modules/audio_processing/aec/echo_cancellation.cc
  modules/audio_processing/aecm/aecm_core.cc
  modules/audio_processing/aecm/aecm_core_c.cc
  modules/audio_processing/aecm/echo_control_mobile.cc
  modules/audio_processing/agc/agc.cc
  modules/audio_processing/agc/agc_manager_direct.cc
  modules/audio_processing/agc/histogram.cc
  modules/audio_processing/agc/legacy/analog_agc.c
  modules/audio_processing/agc/legacy/digital_agc.c
  modules/audio_processing/agc/utility.cc
  modules/audio_processing/audio_buffer.cc
  modules/audio_processing/audio_processing_impl.cc
  modules/audio_processing/beamformer/array_util.cc
  modules/audio_processing/beamformer/covariance_matrix_generator.cc
  modules/audio_processing/beamformer/nonlinear_beamformer.cc
  modules/audio_processing/echo_cancellation_impl.cc
  modules/audio_processing/echo_control_mobile_impl.cc
  modules/audio_processing/gain_control_for_experimental_agc.cc
  modules/audio_processing/gain_control_impl.cc
  modules/audio_processing/high_pass_filter_impl.cc
  modules/audio_processing/intelligibility/intelligibility_enhancer.cc
  modules/audio_processing/intelligibility/intelligibility_utils.cc
  modules/audio_processing/level_estimator_impl.cc
  modules/audio_processing/logging/aec_logging_file_handling.cc
  modules/audio_processing/noise_suppression_impl.cc
  modules/audio_processing/ns/noise_suppression.c
  modules/audio_processing/ns/ns_core.c
  modules/audio_processing/rms_level.cc
  modules/audio_processing/splitting_filter.cc
  modules/audio_processing/three_band_filter_bank.cc
  modules/audio_processing/transient/file_utils.cc
  modules/audio_processing/transient/moving_moments.cc
  modules/audio_processing/transient/transient_detector.cc
  modules/audio_processing/transient/transient_suppressor.cc
  modules/audio_processing/transient/wpd_node.cc
  modules/audio_processing/transient/wpd_tree.cc
  modules/audio_processing/typing_detection.cc
  modules/audio_processing/utility/block_mean_calculator.cc
  modules/audio_processing/utility/delay_estimator.cc
  modules/audio_processing/utility/delay_estimator_wrapper.cc
  modules/audio_processing/vad/gmm.cc
  modules/audio_processing/vad/pitch_based_vad.cc
  modules/audio_processing/vad/pitch_internal.cc
  modules/audio_processing/vad/pole_zero_filter.cc
  modules/audio_processing/vad/standalone_vad.cc
  modules/audio_processing/vad/vad_audio_proc.cc
  modules/audio_processing/vad/vad_circular_buffer.cc
  modules/audio_processing/vad/voice_activity_detector.cc
  modules/audio_processing/voice_detection_impl.cc

  modules/audio_processing/test/test_utils.cc    

  system_wrappers/source/aligned_malloc.cc
  system_wrappers/source/cpu_features.cc
  system_wrappers/source/file_impl.cc
  system_wrappers/source/logging.cc
  system_wrappers/source/metrics_default.cc
  system_wrappers/source/rw_lock.cc
  system_wrappers/source/rw_lock_posix.cc
  system_wrappers/source/trace_impl.cc
  system_wrappers/source/trace_posix.cc
   
)

function(prepend_path var prefix)
  set(listVar "")
  foreach(f ${ARGN})
     list(APPEND listVar "${prefix}/${f}")
  endforeach(f)
  set(${var} "${listVar}" PARENT_SCOPE)
endfunction(prepend_path)

prepend_path(WEBRTC_SRC webrtc/webrtc ${WEBRTC_SRC_})
add_executable(webrtc-bf nonlinear_beamformer_test.cc ${WEBRTC_SRC})
target_link_libraries(webrtc-bf gflags pthread)
SET(CMAKE_CXX_FLAGS_DEBUG "$ENV{CXXFLAGS} -O0 -Wall -g -ggdb")
SET(CMAKE_CXX_FLAGS_RELEASE "$ENV{CXXFLAGS} -O3 -Wall")

为了省事，将相关的文件全部加上了。然后在ubuntu下编译。首先新建build，然后在build目录里执行cmake ..以及make命令：

mkdir build
cmake ..
make

make成功后出现

图片

此时在build目录下生成了可执行文件

图片

然后使用如下命令，

./webrtc-bf -i input.wav -mic_positions "x1 y1 z1 x2 y2 z2" -o output.wav

如果输入的是两通道的语音，那么-mic_positions后面的坐标就是两个麦克风的坐标，多通道的话，以此类推。因为carmix.wav是一个双通道的数据，所以将麦克风的坐标设置为0.02 0 0 -0.02 0 0,坐标原点就是两个麦克风的中心位置。

图片

这样就运行成功了。相关的文件已经上传到github了：https://github.com/ctwgL/webrtc-beamforming

本文就主要介绍了webrtc里的波束模块编译过程，具体的代码细节，技术原理以及处理的效果将会在下篇文章中介绍。

作者：TingweiSpeak 来源：公众号——音频探险记，转载请注明

版权声明：本文内容转自互联网，本文观点仅代表作者本人。本站仅提供信息存储空间服务，所有权归原作者所有。如发现本站有涉嫌抄袭侵权/违法违规的内容， 请发送邮件至1393616908@qq.com 举报，一经查实，本站将立刻删除。

赞 (0)
RTC观主的头像RTC观主认证作者
0 0

相关推荐

发表回复

登录后才能评论