Hi,
I got these strange results:


Source:
WAV 48kHz / 32bit size 1,44 GB

WavPack x6:
998 MB

FLAC 8:
698 MB

-------------------------------
The same source at 24 bit:

WavPack x6:
645 MB

FLAC 8:
654 MB



Is it normal?

Looks like some of the 32 bits were always zero throughout the file.

In foobar2000 settings max. supported bitdepth for FLAC is 24. And I cannot manually encode 32bit files with FLAC. So it seems like FLAC compresses (dithered?) 24-bit wav file.

You are right! It is 24 bit.

You are right.

foobar is doing that, flac.exe never changes the bit depth because that's not lossless.

That's what I tried to say... Fb2k determines that flac supports no more than 24bit and so have to convert resolution from 32 to 24 bit before feeding data to flac.

http://flac.sourceforge.net/format.html#me...lock_streaminfo

"<5> (bits per sample)-1. FLAC supports from 4 to 32 bits per sample. Currently the reference encoder and decoders only support up to 24 bits per sample. "


http://flac.sourceforge.net/format.html#frame_header

Also note here no 32bit is defined 111 is still reserved.

"32" in foobar2000 is 32-bit floating-point. FLAC does only support integer data.

Yet another regression in functionality? In Foobar 0.8 one is able to choose between 32 bits fixed point and float.

No one used that needless functionality, I believe so.

So that's called evolution, as you probably know. Scientists think humans once had a tail ("We thus learn that man is descended from a hairy quadruped furnished with a tail"), so one was able to wag!

I dont see how that actually makes the flac encoder encode 32bit int or float. Nor does it explain if foobars flac decoder can decode 32bit int or float.. Your statement sounds just like internal workings of foobar on other formats rather than anything to do with flac/foobar supporting 32bit.

From my understanding all of kmixer is based on floats so regardless of what you send it the data will be converted to a float for it's own workings. Picking between a 32bit int and float output is only handy when dealing with crappy drivers like Creative release which dont support float in many cases, unless that has been fixed in the last few years.

The discussion is about the Diskwriter/Converter not playback.

FLAC v1.2.1 itself does not encode 32-bits, so Foobar has to downconvert.



BTW, I just checked that Foobar 0.9.5.1 has 8-bit as an option. Do you think it has more use?

I would rate 8-bit as a lot more useful than 32-bit.