So I just hope to clarify about the WAV n FLAC (and all other compressed lossless)

Try not to upsample if you can, I think WASAPI in foobar enables bit perfect playback

Actually, soundcards usually accept only linear PCM encoding, which is also used in .wav files, so every file must be converted to this during playback, even MP3.

FLAC is basically a different encoding for the same data, one which happens to be more space efficient with typical music, at the cost of lower space efficiency with some other signals (encoding white noise with FLAC would likely increase file size).

A 24b 96kHz FLAC file is decoded to the exact same 24b 96kHz PCM signal which was used to create this file. A 24b 96kHz MP3 file would get decoded to a slightly different 24b 96kHz PCM, because it doesn't preserve all information about original signal in order to achieve better size reduction.
And you were right, all lossless files decode exactly to their original PCMs.

24(for the bit depth)*96(kHz)*2(channels) = 4608kbps. Still PCM data.

 

Bitrate from an uncompressed file comes from the amount of channels, the bit depth, and the sample rate as shown above. Bitrate for a lossless compressed file is only an indication of how well it was compressed.

There's also DTS DSD, which is used on SACDs I believe.

 

WAV requires no decoding, it's just PCM data with a wrapper. Yes, it should be faster than FLAC/ALAC but buffers totally negate this. It might take some extra milliseconds to start your file.

 

Even if the buffer couldn't keep up the result would be dropouts, not changes in sound quality.

Edited by chewy4 - 2/15/13 at 10:06am

Decoding FLAC is also several hundred times faster than real time on current CPUs. The various background processes running on a multi-tasking OS could very well consume more CPU time.

SACDs use a process called Direct Stream Digital and Direct Stream Transmission, and are on the disc at a different depth than the standard Red Book info.  It has the capability of multichannel audio, but has found most application in stereo.  Because it uses different depth layers on the disc, it takes a special player to get the data off the disc, it's not a simple software decoder.

 

DTS (Digital Theater Systems) is a company that produces their own audio coding systems, primarily for multichannel audio, but the actual channel count is almost unlimited, based only on the available bandwidth.  There are many flavors of DTS, the most common of which appears in motion picture sound in theaters, and on DVDs, including audio-only DVDs.  Some feel it is more transparent than Dolby Digital. You'll also find their DTS Master Audio format, which has the capability of unlimited channels of 24/192 lossless.  Software decoders are possible. 

Thanks for the correction, DSD is what I was thinking of.

FLAC is very fast at decoding. Foobar2000 shows about 0.03% CPU usage increase when playing FLAC instead of WAV on my computer. I could decode a couple hundred or maybe thousand FLAC files simultaneously and it would still be faster than realtime.

 

The claims that WAV sounds better is audiophile nonsense.