Hey all. I have ripped a large portion of my cd collection to wavpack using 16 bit and EAC. Converting from SHN to wavpack in foobar I recently found out how to encode as 24 bit. Probably a stupid question, but I never found out how to do this in EAC and is this difference important (i.e. important enough to re-encode my CDs)? Same issue with my FLAC files as well...
CDs are stored as 16-bit, so the additional 8 bits will all be zero. Nothing gained.
(HDCDs could be post-processed.)
CDs are stored as 16-bit, so the additional 8 bits will all be zero. Nothing gained.
(HDCDs could be post-processed.)
Cool thanks. So it is important that my lossy codecs such as MP3 seem to be encoded as 24 bit but not CDs, correct? And what do you mean by post-processed?...
CDs are stored as 16-bit, so the additional 8 bits will all be zero. Nothing gained.
(HDCDs could be post-processed.)
Cool thanks. So it is important that my lossy codecs such as MP3 seem to be encoded as 24 bit but not CDs, correct? And what do you mean by post-processed?...
I don't quite understand your question on mp3s. Anyway, mp3s should be kept the way you have them, not transcoded, as converting an mp3 to an mp3 of same size does reduce quality.
"HDCD" discs are CDs in physical format, but the 16th bit contains information saying "you should run it through this digital effect". That is, it can be detected from the sound file afterwards, not only when ripped. (If you really care. Ignoring this would lead to the same result as playing a HDCD through a CD player which does not support it.)
Just that all of the mp3 files that I have seem to be 24 bit. Don't think I have any that are 16 bit and most of them I didn't rip personally. So with HDCDs with that 16th bit for effects, you would want 24 bit rips correct?
You seem confused. Mp3s are not 16-bit or 24-bit. The data inside them is 32-bit float, regardless of the bit depth of the source used to encode them. Unlike the bit depth of the source files which are the resolution of the individual samples in the time domain, these 32-bit float numbers represent coefficients for frequency bands.
with HDCDs with that 16th bit for effects, you would want 24 bit rips correct?
You can
rip HDCDs to 16 bit without any loss whatsoever. Then:
If your player supports HDCD, it will accommodate the 16th bit coding.
If you wish to do that yourself, then there's this hdcd.exe application which will run through your 16 bit FLAC file, detect HDCD if applicable (and otherwise leave file as-is), do the processing and output a file of 24 bits. But I've learned -- a bit too late, unfortunately -- that you need not re-rip anything, nor do you need to rip to anything more than 16 bits, 44.1 sampling rate.
I don't know if the OP uses HDCDs, but I'll supplement Porcus's post by mentioning that kode54's HDCD decoder for foobar2000 (http://www.hydrogenaudio.org/forums/index.php?showtopic=79427) was released recently.
Just that all of the mp3 files that I have seem to be 24 bit. Don't think I have any that are 16 bit and most of them I didn't rip personally.
MP3s don't have a number of bits associated with them, so you're either misunderstanding something else or using crappy software that is confused about your files.
I use XBMC running on a softmodded xbox and when you get info about the file, it lists the mp3 files as 24 bit. Hence the confusion I guess...
I guess the decoder is reporting the number of bits it outputs. In terms of the accuracy of the raw data, it's as meaningless as the number of decimal places supported (output) by a calculator.
e.g. 2.1/3.1 = lots of decimal places (even though the input only had 1 decimal place)
An mp3 contains a lot of floating point data, that's both far less accurate than the input, but creates far more (infinite?) places/bits in the output.
The only value to this is that having more bits reduces the amount of additional noise you add at the output - but that noise is already far far less than has been added by the mp3 encoding process itself (where lots and lots of noise = imprecision is introduced).
Cheers,
David.
But anyway, to the OP: Never transcode lossy files for quality.