Looking for the previous docs site? Visit the legacy docs homepage. (opens in a new tab)
Agora Docs

Custom audio source

Updated

Integrate a custom video or audio capture into your client

The default audio module of Video SDK meets the need of using basic audio functions in your game. For adding advanced audio functions, Video SDK supports using custom audio sources and custom audio rendering modules.

Video SDK uses the basic audio module on the device your game runs on by default. However, there are certain use-cases where you want to integrate a custom audio source into your game, such as:

  • Your game has its own audio module.
  • You need to process the captured audio with a pre-processing library for audio enhancement.
  • You need flexible device resource allocation to avoid conflicts with other services.

This page shows you how to capture and render audio from custom sources.

Understand the tech

To set an external audio source, you configure the Agora Engine before joining a channel. To manage the capture and processing of audio frames, you use methods from outside the Video SDK that are specific to your custom source. Video SDK enables you to push processed audio data to the subscribers in a channel.

Capture custom audio

The following figure illustrates the process of custom audio capture.

  • You implement the capture module using external methods provided by the SDK.

  • You call pushAudioFrame to send the captured audio frames to the SDK.

Render custom audio

The following figure illustrates the process of custom audio rendering.

  • You implement the rendering module using external methods provided by the SDK.

  • You call pullAudioFrame to retrieve the audio data sent by remote users.

Prerequisites

Ensure that you have implemented the SDK quickstart in your project.

Implementation

This section shows you how to implement custom audio capture and render audio from an external source using the SDK.

Initialize MediaEngine

Before implementing custom audio features, obtain the MediaEngine interface from the initialized RtcEngine:

void InitAgoraEngine(FString APP_ID, FString TOKEN, FString CHANNEL_NAME)
{
  // Initialize RtcEngine context and event handler
  agora::rtc::RtcEngineContext RtcEngineContext;
  UserRtcEventHandler = MakeShared<FUserRtcEventHandler>(this);
  std::string StdStrAppId = TCHAR_TO_UTF8(*APP_ID);
  RtcEngineContext.appId = StdStrAppId.c_str();
  RtcEngineContext.eventHandler = UserRtcEventHandler.Get();
  RtcEngineContext.channelProfile = agora::CHANNEL_PROFILE_TYPE::CHANNEL_PROFILE_LIVE_BROADCASTING;

  // Initialize the RtcEngine
  int ret = AgoraUERtcEngine::Get()->initialize(RtcEngineContext);

  // Query for the MediaEngine interface
  AgoraUERtcEngine::Get()->queryInterface(AGORA_IID_MEDIA_ENGINE, (void**)&MediaEngine);
}

The MediaEngine is not created directly. Instead, it's obtained as an interface from the initialized RtcEngine using the queryInterface method with AGORA_IID_MEDIA_ENGINE.

Custom audio capture

Refer to the following call sequence diagram to implement custom audio capture in your app:

Custom audio capture process

Follow these steps to implement custom audio capture in your project:

  1. Enable and configure custom audio source Before calling joinChannel to join the channel, call setExternalAudioSource to enable and configure custom audio capture.
void SetExternalAudioSource()
{
  // Specify the custom audio source
  int ret = MediaEngine->setExternalAudioSource(true, SAMPLE_RATE, CHANNEL, 1);
  UBFL_Logger::Print(FString::Printf(TEXT("%s ret %d"), *FString(FUNCTION_MACRO), ret), LogMsgViewPtr);
}

void JoinChannel()
{
  // Enable audio and set client role
  AgoraUERtcEngine::Get()->enableAudio();
  AgoraUERtcEngine::Get()->setClientRole(CLIENT_ROLE_BROADCASTER);

  // Join the channel
  int ret = AgoraUERtcEngine::Get()->joinChannel(TCHAR_TO_UTF8(*Token), TCHAR_TO_UTF8(*ChannelName), "", 0);
}

  1. Implement audio capture and processing Use methods outside the SDK to implement audio capture and processing yourself.

  2. Send audio frames to SDK Call pushAudioFrame to send the captured audio frames to the SDK for later use.

  3. Load audio data from a file and start a thread to push the frames:

    void StartPushAudio()
{
  FString LoadDir = FPaths::ProjectContentDir() / TEXT("Audio/Agora.io-Interactions.wav");
  TArray<uint8> AudioData;
  FFileHelper::LoadFileToArray(AudioData, *LoadDir, 0);
  Runnable = new FAgoraCaptureRunnable(MediaEngine, AudioData.GetData(), AudioData.Num() * sizeof(uint8));
  FRunnableThread* RunnableThread = FRunnableThread::Create(Runnable, TEXT("AgoraUE-UserThread"));
}

  4. This class reads audio data at 10ms intervals, converts it into AudioFrame format, and sends it to the Agora SDK using pushAudioFrame.

    class FAgoraCaptureRunnable : public FRunnable
{
public:
  FAgoraCaptureRunnable(agora::media::IMediaEngine* MediaEngine, const uint8* audioData, int dataLength)
    : MediaEngine(MediaEngine), dataLength(dataLength)
  {
    this->audioData = new uint8[dataLength];
    FMemory::Memcpy(this->audioData, audioData, dataLength * sizeof(uint8));
  }

  virtual uint32 Run() override
  {
    auto tic = getTimeStamp();
    bStopThread = false;
    const uint8* Ptr = reinterpret_cast<const uint8*>(audioData);

    while (!bStopThread)
    {
      if (MediaEngine == nullptr) break;

      auto toc = getTimeStamp();
      if ((toc - tic) >= 10) // Push every 10ms
      {
        if (dataLength <= 0) break;

        if (sendByte == nullptr)
        {
          sendByte = FMemory::Malloc(SAMPLE_RATE / PUSH_FREQ_PER_SEC *
                      agora::rtc::BYTES_PER_SAMPLE::TWO_BYTES_PER_SAMPLE * CHANNEL);
        }

        FMemory::Memcpy(sendByte, Ptr, SAMPLE_RATE / PUSH_FREQ_PER_SEC *
               agora::rtc::BYTES_PER_SAMPLE::TWO_BYTES_PER_SAMPLE * CHANNEL);

        // Prepare audio frame
        agora::media::IAudioFrameObserverBase::AudioFrame externalAudioFrame;
        externalAudioFrame.bytesPerSample = agora::rtc::BYTES_PER_SAMPLE::TWO_BYTES_PER_SAMPLE;
        externalAudioFrame.type = agora::media::IAudioFrameObserver::FRAME_TYPE_PCM16;
        externalAudioFrame.samplesPerChannel = SAMPLE_RATE / PUSH_FREQ_PER_SEC;
        externalAudioFrame.samplesPerSec = SAMPLE_RATE;
        externalAudioFrame.channels = CHANNEL;
        externalAudioFrame.buffer = sendByte;
        externalAudioFrame.renderTimeMs = 10;

        // Push audio frame to SDK
        agora::rtc::track_id_t trackId = 0;
        int ret = MediaEngine->pushAudioFrame(&externalAudioFrame, trackId);

        Ptr += SAMPLE_RATE / PUSH_FREQ_PER_SEC * 2 * CHANNEL;
        dataLength -= SAMPLE_RATE / PUSH_FREQ_PER_SEC * 2 * CHANNEL;
        tic = toc;
      }
      FPlatformProcess::Sleep(0.001f);
    }
    return 0;
  }

private:
  agora::media::IMediaEngine* MediaEngine = nullptr;
  uint8* audioData = nullptr;
  void* sendByte = nullptr;
  int dataLength = 0;
  bool bStopThread = false;

  // Audio configuration
  const int CHANNEL = 2;
  const int SAMPLE_RATE = 48000;
  const int PUSH_FREQ_PER_SEC = 100;
};

Custom audio rendering

This section shows you how to implement custom audio rendering. Refer to the following call sequence diagram to implement custom audio rendering in your game:

Custom audio rendering workflow

To implement custom audio rendering, use the following methods:

  1. Enable and configure custom audio sink Before calling joinChannel to join the channel, call setExternalAudioSink to enable and configure custom audio rendering.
void SetExternalAudioSink()
{
  // Enable custom audio rendering
  // Sample rate (Hz) can be set to 16000, 32000, 44100 or 48000
  // Number of channels can be set to 1 or 2
  int CHANNEL = 1;
  int SAMPLE_RATE = 44100;
  int ret = MediaEngine->setExternalAudioSink(true, SAMPLE_RATE, CHANNEL);
}
  1. Pull and render remote audio data To pull audio frames for custom rendering, call pullAudioFrame in a dedicated thread. After pulling the frame, send the audio buffer to your custom sound player (such as a USoundWaveProcedural wrapper like AgoraSoundWaveProcedural).
class FAgoraRenderRunnable : public FRunnable
{
public:
  FAgoraRenderRunnable(agora::media::IMediaEngine* InMediaEngine, UAgoraSoundWaveProcedural* InAgoraSoundWaveProcedural)
    : MediaEngine(InMediaEngine), AgoraSoundWaveProcedural(InAgoraSoundWaveProcedural)
  {
  }

  virtual uint32 Run() override
  {
    constexpr int PUSH_FREQ_PER_SEC = 100;
    constexpr int SAMPLE_RATE = 44100;
    constexpr int CHANNEL = 1;

    auto tic = getTimeStamp();
    bStopThread = false;

    // Prepare the external audio frame
    agora::media::IAudioFrameObserverBase::AudioFrame externalAudioFrame;
    externalAudioFrame.avsync_type = 0; // Reserved parameter
    externalAudioFrame.bytesPerSample = agora::rtc::BYTES_PER_SAMPLE::TWO_BYTES_PER_SAMPLE;
    externalAudioFrame.type = agora::media::IAudioFrameObserver::FRAME_TYPE_PCM16;
    externalAudioFrame.samplesPerChannel = SAMPLE_RATE / PUSH_FREQ_PER_SEC;
    externalAudioFrame.samplesPerSec = SAMPLE_RATE;
    externalAudioFrame.channels = CHANNEL;
    externalAudioFrame.renderTimeMs = 10;
    externalAudioFrame.buffer = FMemory::Malloc(
      externalAudioFrame.samplesPerChannel *
      externalAudioFrame.bytesPerSample *
      externalAudioFrame.channels
    );

    while (!bStopThread)
    {
      if (MediaEngine == nullptr) break;

      auto toc = getTimeStamp();
      if ((toc - tic) >= 10) // Pull every 10 ms
      {
        tic = getTimeStamp();

        // Pull remote audio data
        int ret = MediaEngine->pullAudioFrame(&externalAudioFrame);
        if (ret == 0 && AgoraSoundWaveProcedural)
        {
          // Send the frame to custom renderer (e.g., USoundWaveProcedural wrapper)
          AgoraSoundWaveProcedural->AddToFrames(externalAudioFrame);
        }
        else if (ret != 0)
        {
          FPlatformProcess::Sleep(0.01f); // Sleep 10ms if no data
          continue;
        }
      }
      FPlatformProcess::Sleep(0.001f); // Yield to CPU
    }

    // Free audio buffer after loop
    FMemory::Free(externalAudioFrame.buffer);
    return 0;
  }

private:
  agora::media::IMediaEngine* MediaEngine = nullptr;
  UAgoraSoundWaveProcedural* AgoraSoundWaveProcedural = nullptr;
  bool bStopThread = false;
};
  • After joining the channel, call pullAudioFrame to get the audio data sent by the remote user.
  • Use your own audio renderer to process the audio data and then play the rendered data.

Using raw audio data callback

This section explains how to implement custom audio rendering.

To retrieve audio data for playback, implement collection and processing of raw audio data. Refer to Raw audio processing.

Follow these steps to call the raw audio data API in your project for custom audio rendering:

  1. Retrieve audio data for playback using the onRecordAudioFrame, onPlaybackAudioFrame, onMixedAudioFrame, or onPlaybackAudioFrameBeforeMixing callback.

  2. Independently render and play the audio data.

Reference

This section explains how to implement different sound effects and audio mixing in your game, covering essential steps and code snippets.

Sample projects

Agora provides the following open-source sample projects for audio self-capture and audio self-rendering for your reference:

API reference

Contact Us

+1 (408) 879-5885

2804 Mission College Blvd.Santa Clara, CA, USA 95054

AgoraCopyright © 2026 AgoraAll rights reserved