Is it possible to Intercept and postprocess all audio streams coming from any app or process running on Windows device?

I did some tests with WASAPI (loopback mode) and I could modify sound from Youtube videos (but the resulting sound was garbage, as I don't know how to add a correct filter from audio data (KSDATAFORMAT_SUBTYPE_IEEE_FLOAT subformat))

@Castorix31 Could you please help me with some boilerplate code or point me to useful resources .

In your test, I hope actual audio from youtube was not getting played and only garbage audio data is played in output .

I'm not sure if it is not played, then garbage added above , which would not be good (I tried to remove completely the sound with an empty buffer but I could not...)

I used code from https://video2.skills-academy.com/en-us/windows/win32/coreaudio/loopback-recording

then at same time a Render client (I generated the following function from ChatGPT) , then GetBuffer to replace audio data with new modified audio data)

HRESULT InitializeRenderClient(IMMDevice** ppDevice, IAudioClient** ppAudioClient, IAudioRenderClient** ppRenderClient, WAVEFORMATEX* pwfx)
{
    HRESULT hr;
    IMMDeviceEnumerator* pEnumerator = nullptr;
    hr = CoCreateInstance(__uuidof(MMDeviceEnumerator), nullptr, CLSCTX_ALL, IID_PPV_ARGS(&pEnumerator));
    if (FAILED(hr)) return hr;

    // Get the default output audio device (or specify another device)
    hr = pEnumerator->GetDefaultAudioEndpoint(eRender, eConsole, ppDevice);
    if (FAILED(hr)) return hr;

    // Activate the audio client for the render device
    hr = (*ppDevice)->Activate(__uuidof(IAudioClient), CLSCTX_ALL, nullptr, (void**)ppAudioClient);
    if (FAILED(hr)) return hr;

    // Initialize the audio client for rendering (output)
    hr = (*ppAudioClient)->Initialize(AUDCLNT_SHAREMODE_SHARED, 0, 10000000, 0, pwfx, nullptr);
    if (FAILED(hr)) return hr;

    // Get the render client
    hr = (*ppAudioClient)->GetService(IID_PPV_ARGS(ppRenderClient));
    return hr;
}