Audio Sample Manipulation

Question

Hello, I'm developing an audio application and don't know how to manipulate the data that comes from IAudioCaptureClient ::GetBuffer().

I need to manipulate the sample to change the wave form and produce some bass guitar effects like overdrive, chorus and delay.

I inspect the data from WAVEFORMATEX structure to try to discovery something but I can't.

The code to print the data from WAVEFORMATEX is:

printf("Samples per second: %d\n", pwfx->nSamplesPerSec);
printf("Channels: %d\n", pwfx->nChannels);
printf("Size of extra information: %d\n", pwfx->cbSize);
printf("Block alignment in bytes: %d\n", pwfx->nBlockAlign);
printf("Format type code: %d\n", pwfx->wFormatTag);
printf("Bits per sample: %d\n", pwfx->wBitsPerSample);
printf("Required average data tranfer rate: %d\n", pwfx->nAvgBytesPerSec);

And the output is:

Samples per second: 44100
Channels: 2
Size of extra information: 22
Block alignment in bytes: 8
Format type code: 65534  //this code stands for WAVE_FORMAT_EXTENSIBLE
Bits per sample: 32
Required average data tranfer rate: 352800

Someone can help me with this?

Thank you

Answer 1

When the format type code is WAVE_FORMAT_EXTENSIBLE, cast pwfx to a WAVEFORMATEXTENSIBLE.  Then inspect the SubFormat member.  See http://blogs.msdn.com/b/matthew_van_eerde/archive/2011/09/09/how-to-validate-and-log-a-waveformatex.aspx

I suspect in this case it will be equal to KSDATAFORMAT_SUBTYPE_IEEE_FLOAT.  This means that each sample is a four-byte floating-point integer of type "float".

To back up a step, where are you getting the data from?  Is it coming from a recording device like the line in or microphone, or are you doing WASAPI loopback capture from a playback device like the speakers or the headphones?

If you're pulling it from a recording device, you can do what you like with the data and then hand it off to somewhere.

If you're pulling it from a playback device using WASAPI loopback capture, then be aware that WASAPI loopback capture is a read-only tap; modifications you make to the data will not be applied to what goes out of the speakers.

---

Matthew van Eerde

Answer 2

Matthew, i will make the cast after because here I don't have my source code.

I'm getting the data from microphone and want to modify the resulting sound.

Guilherme

Answer 3

OK.

Assuming that the audio is indeed in float format (and the SubFormat will tell you this) you can modify the data in a given packet something like this:

BYTE *pData = NULL;
UINT32 nFrames = 0;
pAudioCaptureClient::GetBuffer(&pData, &nFrames, ...);
float *pFloatData = reinterpret_cast<float*>(pData);
for (UINT32 f = 0; f < nFrames; f++) {
    for (WORD c = 0; c < pWfx->nChannels; c++) {
        // this is the sample corresponding to the c'th channel in frame f
        float sample = pFloatData[f * pWfx->nChannels + c];
        // do your processing here
    }
}

---

Matthew van Eerde

Answer 4

The format is correct Matthew.

Now I must create a temporary buffer to store data before processing.

I'm just starting this project and guess that some problems will arise.

Thanks for your help.