I have a Voice Call implementation that uses the Oboe library for low latency, and it works fine on the C++ layer. On the Java side, when I don't select Bluetooth SCO, everything functions well with the Bluetooth headset buttons; I can start and stop the call using the headset button without any issues, and the media session events work correctly. However, the downside is that it does not use the headset microphone; it defaults to the phone's microphone.

To resolve this, I used startBluetoothScoOn, which allows the headset microphone to work properly. Unfortunately, this causes the headset button actions to stop functioning.

Option1. Headset button: OK | Headset speaker: OK | Routing to headset mic: FAIL

audioManager().setMode(AudioManager.MODE_NORMAL);
audioManager().setSpeakerphoneOn(false)

Option2. Headset button: FAIL | Headset speaker: OK | Routing to headset mic: OK

audioManager().setMode(AudioManager.MODE_NORMAL);
audioManager().setSpeakerphoneOn(false);
audioManager().setBluetoothScoOn(true);
audioManager().startBluetoothSco();

Oboe Details:

oboe::AudioStreamBuilder *
AndroidAudioDeviceManager::setupCommonStreamParameters(
    oboe::AudioStreamBuilder *builder) {
  
  builder->setAudioApi(mAudioApi)
      ->setFormat(mFormat)
      ->setFormatConversionAllowed(true)
      ->setSharingMode(oboe::SharingMode::Exclusive)
      ->setUsage(oboe::Usage::VoiceCommunication)
      ->setContentType(oboe::ContentType::Speech)
      ->setPerformanceMode(oboe::PerformanceMode::LowLatency);
  return builder;

With the option 1, the MediaSession capture the headset button events, using this:

private void startMediaSession () {
    
    if (_mediaSession == null) {
        _mediaSession = new MediaSession(App.getContext(), TAG);
    } else return;

    _mediaSession.setCallback(new MediaSession.Callback() {

        @Override
        public boolean onMediaButtonEvent(Intent mediaButtonIntent) {
            KeyEvent keyEvent = mediaButtonIntent.getParcelableExtra(Intent.EXTRA_KEY_EVENT);
            // I process the events here and works fine with Option 1.
    }, _handler);
    _mediaSession.setActive(true);
}

With Option 2, when you perform the headset button action, we do not receive MediaSession events. Instead, the entire audio layer in C++ is restarted, causing the audio streams to restart as well. This makes handling the button action quite challenging.

In contrast, with Option 1, the headset action button works smoothly and quickly, without causing any audio stream restarts in the C++ layer. However, the audio routing to the headset microphone does not occur.

NOTE: When using a USB-C headset and its buttons, everything works perfectly with no issues, and the same code is applied. The audio routing between devices is also very clear; for example, switching from the USB-C headset to the built-in speaker or responding to actions with the USB-C headset button functions seamlessly.

All of this is implemented using target SDK 35.

I believe there is a way to resolve this issue without forcing changes in the C++ layer. Perhaps someone has insights on how to achieve this?