native webrtc支持切换音频采集设备和获取裸流

https://www.yuque.com/caokunchao/rtendq/oq8w3qgs3g59whru

前言

版本webrtc m96
1、修改webrtc m96代码，向外提供一个adm指针的接口出来
2、外部来获取指针进行设备的选择
3、外部获取音频裸流，麦克风或者扬声器的数据

修改webrtc代码

1、修改H:\webrtc\webrtc-checkout\webrtc\api\peer_connection_interface.h，PeerConnectionFactoryInterface类
添加接口

class AudioDeviceModule;
virtual rtc::scoped_refptr<AudioDeviceModule> GetAdmPtr() = 0;

2、修改H:\webrtc\webrtc-checkout\webrtc\pc\peer_connection_factory.h，
PeerConnectionFactory类，该类继承PeerConnectionFactoryInterface，实现GetAdmPtr()接口

#include "modules/audio_device/include/audio_device.h"
#include "media/base/media_engine.h"rtc::scoped_refptr<AudioDeviceModule> GetAdmPtr() override {return context_->channel_manager()->media_engine()->voice().GetAdm();}

3、修改代理，H:\webrtc\webrtc-checkout\webrtc\pc\peer_connection_factory_proxy.h

#include "modules/audio_device/include/audio_device.h"在 BEGIN_PROXY_MAP(PeerConnectionFactory) 下面添加代理方法
***
PROXY_METHOD0(rtc::scoped_refptr<AudioDeviceModule>,GetAdmPtr)
***
END_PROXY_MAP(PeerConnectionFactory)

4、修改H:\webrtc\webrtc-checkout\webrtc\media\base\media_engine.h，VoiceEngineInterface类
添加接口

 virtual rtc::scoped_refptr <webrtc::AudioDeviceModule> GetAdm() = 0;

5、修改H:\webrtc\webrtc-checkout\webrtc\media\engine\webrtc_voice_engine.h，WebRtcVoiceEngine类
实现接口

rtc::scoped_refptr<webrtc::AudioDeviceModule> GetAdm() override { return adm_; }

外部切换设备

看我的krtcsdk源码

void MicImpl::Start() {RTC_LOG(LS_INFO) << "MicImpl Start call";KRTCGlobal::Instance()->worker_thread()->PostTask(webrtc::ToQueuedTask([=]() {RTC_LOG(LS_INFO) << "MicImpl Start PostTask";KRTCError err = KRTCError::kNoErr;do {// 1. 如果麦克风已经启动采集，直接停止if (has_start_) {RTC_LOG(LS_WARNING) << "mic already start, mic_id: " << mic_id_;break;}// 2. 直接从webrtc获取adm模块指针rtc::scoped_refptr<webrtc::AudioDeviceModule> audio_device =KRTCGlobal::Instance()->push_peer_connection_factory()->GetAdmPtr();// 3. 检查系统是否存在麦克风设备int total = audio_device->RecordingDevices();if (total <= 0) {RTC_LOG(LS_WARNING) << "no audio device";err = KRTCError::kNoAudioDeviceErr;break;}// 4. 检查关联的mic_id是否能够在系统设备中找到int device_index = -1;for (int i = 0; i < total; ++i) {char name[128];char guid[128];audio_device->RecordingDeviceName(i, name, guid);if (0 == strcmp(guid, mic_id_.c_str())) {device_index = i;break;}}if (device_index <= -1) {RTC_LOG(LS_WARNING) << "audio device not found, mic_id: " << mic_id_;err = KRTCError::kAudioNotFoundErr;break;}// 5. 设置启用的麦克风设备if (audio_device->SetRecordingDevice(device_index)) {RTC_LOG(LS_WARNING) << "SetRecordingDevice failed, mic_id: " << mic_id_;err = KRTCError::kAudioSetRecordingDeviceErr;break;}// 6. 设置为立体声采集audio_device->SetStereoRecording(true);// 7. 初始化麦克风if (audio_device->InitRecording() || !audio_device->RecordingIsInitialized()) {RTC_LOG(LS_WARNING) << "InitRecording failed, mic_id: " << mic_id_;err = KRTCError::kAudioInitRecordingErr;break;}bool ok = false;audio_device->PlayoutIsAvailable(&ok);if (!ok) {RTC_LOG(LS_WARNING) << "PlayoutIsAvailable failed, mic_id: " << mic_id_;err = KRTCError::kAudioInitRecordingErr;break;}int32_t ret = audio_device->InitPlayout();if (audio_device->StartPlayout()) {RTC_LOG(LS_WARNING) << "StartPlayout failed!!!";err = KRTCError::kAudioStartRecordingErr;break;}// 8. 启动麦克风采集if (audio_device->StartRecording()) {RTC_LOG(LS_WARNING) << "StartRecording failed, mic_id: " << mic_id_;err = KRTCError::kAudioStartRecordingErr;break;}has_start_ = true;} while (0);if (err == KRTCError::kNoErr) {if (KRTCGlobal::Instance()->engine_observer()) {KRTCGlobal::Instance()->engine_observer()->OnAudioSourceSuccess();}}else {if (KRTCGlobal::Instance()->engine_observer()) {KRTCGlobal::Instance()->engine_observer()->OnAudioSourceFailed(err);}}})); 
}

这里音频audio_device->StartRecording之前，还必须加上audio_device->StartPlayout()否则会报错
(audio_device_core_win.cc:2351): Playout must be started before recording when using the built-in AEC
外部获取音频裸流
1、添加ADMDataObserver，继承自webrtc::AudioDeviceDataObserver

class ADMDataObserver : public webrtc::AudioDeviceDataObserver {
private:virtual void OnCaptureData(const void* audio_samples,const size_t num_samples,const size_t bytes_per_sample,const size_t num_channels,const uint32_t samples_per_sec) override {}virtual void OnRenderData(const void* audio_samples,const size_t num_samples,const size_t bytes_per_sample,const size_t num_channels,const uint32_t samples_per_sec) override {}};

OnCaptureData 音频采集麦克风数据，OnRenderData需要播放的扬声器数据。。
2、创建webrtc::AudioDeviceModule

rtc::scoped_refptr<webrtc::AudioDeviceModule> audio_device_;worker_thread_->Invoke<void>(RTC_FROM_HERE, [=]() {audio_device_ = webrtc::AudioDeviceModule::Create(webrtc::AudioDeviceModule::kPlatformDefaultAudio,task_queue_factory_.get());audio_device_ = webrtc::CreateAudioDeviceWithDataObserver(audio_device_, std::make_unique<ADMDataObserver>());audio_device_->Init();});

3、将audio_device_传入到webrtc::CreatePeerConnectionFactory即可。。

webrtc::CreatePeerConnectionFactory(network_thread_.get(), /* network_thread */worker_thread_.get(), /* worker_thread */signaling_thread_.get(),  /* signaling_thread */audio_device_,  /* default_adm */******

参考资料

https://blog.csdn.net/qq_22658119/article/details/117664188
https://blog.csdn.net/weixin_39343678/article/details/99948451