wav isn't always raw PCM though. Make sure to check the RIFF to see if you need to decode it first.

Yeah, I always check WAVEFORMATEX[TENSIBLE] structure and not blindly assume it's PCM. Made that mistake before, will not do it again.

Anyway, turns out the format is IEEE float. Damn! Looks like I've got some more work to do.

Shogun.

Is it a problem to work with float? Bah, I'll ask in the chat.

I don't know if it works roughly the same as PCM or what, only with 32-bit floating point precision. And what would the maximum value be for a sound sample of this format?

Shogun.

EDIT: I came across this excellent article that explains audio conversion(s), mixing and PCM/IEEE float formatting fairly extensively. http://www.codeproject.com/Articles/501521/How-to-convert-between-most-audio-formats-in-NET

Thanks for the responses.

Yeah, also came to that conclusion.

Now I have one problem, whenever I try to read a float value, I always get this value "-nan(ind)". I tried googling this, but no results came up (mostly garbage). This is my code:

        void* riff = malloc( size );
	WAVEFORMATEXTENSIBLE* wfx = NULL;
	BYTE* pcm = NULL;
	DWORD dwPcmSize;

	fread( riff, size, 1, fp );

	int result = UnpackWAVDataEx( riff, &wfx, &pcm, &dwPcmSize );

	if( !result )
	{
		printf( "Could not read .wav fle!\n" );
	}
	else if( wfx->Format.nChannels != 4 )
	{
		printf( "All wav files are expected to have 4 channels...\n" );
	}
	else if( wfx->SubFormat == KSDATAFORMAT_SUBTYPE_PCM )
	{
		printf( "PCM formatted .wav file detected...\n" );

                /* TODO: If necessary */
	}
	else if( wfx->SubFormat == KSDATAFORMAT_SUBTYPE_IEEE_FLOAT )
	{
		printf( "IEEE float formatted .wav file detected...\n" );

		float* soundbytes = (float*) pcm;
		
		float ch1 = 0.0f;
		float ch2 = 0.0f;
		float ch3 = 0.0f;
		float ch4 = 0.0f;

		for( DWORD i = 0; i < dwPcmSize/4; i += 4 )
		{
			ch1 += GetDB(soundbytes[i+0]);
			ch2 += GetDB(soundbytes[i+1]);
			ch3 += GetDB(soundbytes[i+2]);
			ch4 += GetDB(soundbytes[i+3]);
		}

		float avg[4] = { ch1/4.0f, ch2/4.0f, ch3/4.0f, ch4/4.0f };

		FILE* fout = fopen( "out.txt", "w" );
		fprintf( fout, "Channel 1: %f\nChannel 2: %f\nChannel 3: %f\nChannel 4: %f\n", avg[0], avg[1], avg[2], avg[3] );
		fclose(fout);
	}
	else
	{
		printf( "This .wav format is not supported...\n" );
	}

Am I doing this right? Or do I have to do a conversion of DWORD to float every 4 bytes?

Shogun.

EDIT: Found the problem, it has to do with the way I calculate sound levels based on this algorithm:

dB = 20 * log10(amplitude)

http://stackoverflow.com/questions/2445756/how-can-i-calculate-audio-db-level

The value can't be negative or else I'll get a bogus result. Fixed it.