|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
66.169.245.112
I spend all my time at the Vinyl Asylum, but am posting here because I need some advice from a pro! I am only a serious amateur.
I've been doing vinyl restoration for years, most of it on my approx. 500 LPs. Last year I underwent a MAJOR bigtime upgrade to my vinyl front end and started the digitization of my vinyl all over again. I believe this will be the last time for that, as all of my equipment is of a very high grade now.
I use Waves plug-ins in my editor exclusively. My usual procedure is running the Restoration Bundle, followed by L1 Ultra-Maximizer, which I use to limit a few peaks and to dither. I then downsample to 44.1, then split and save to 16bit wave file. For those who are curious, the splitting places each track into a block of samples which is divisible by 588 samples according to Redbook, which writes the information into 1/75th of a second blocks. (44,100 / 588 = 1/75th.) All this is just saying the track markers are adjusted minisculely to make a CD friendly wave file.
Regarding dither, it is usually agreed that it should be the very last step in a mastering process. But, as noted above, I have always dithered an 88.2kHz/24bit file as its last step, then downsample, split, and save.
It recently occurred to me that this method could make the dither ineffective, since rendering to 16bits after dithering the 24bit file does not put the dither as the last step. I am thinking the dither existed in the last few bits of the 24bit file and gets truncated out of the file when I convert it to 16bit? I know dither is a subtle issue to begin with, but would appreciate any advice/info I can get.
Follow Ups:
.Why would spend the money to upgrade your vinyl set up, go to all of the trouble to digitize all your records again, and not save them as higher resolution like 24/192 which is much better sounding than down sampling to CD quality.
Your front end is of MAJOR quality which you severely compromise by down sampling.
I don't get it.
.
Edits: 09/07/11
Your suspicions are correct, the dither step is not in the correct place, it certainly is not last. "Dither last" is also not always the most accurate rule to follow. A better rule for dither, is dither after any reduction in word length. All audio processing increases the word length.
Are you recording at 24 bit to begin with? If not, the work flow is not technically correct.
You should dither AFTER the sample rate conversion, because any processing done to audio will increase the word length. Currently when you resample your 88/16 file to 44, it becomes 44/24, not 44/16. You can see this happen with any bitmeter plugin.
This is what I'd suggest, based on the fact that a software resample process may not be very high quality either. Avoid a resample process if you can, those processes can be most destructive to audio quality. Also, when using a peak limiter like the L1, I'd recommend setting the highest peaks for maximum of -2 or -3 dBFS to avoid distorting DAC's on playback (see TC Electronic white paper on DAC overload http://www.tcelectronic.com/media/nielsen_lund_2003_overload.pdf)
Record your LP in at 44/24. Process with whatever you want. Dither to 44/16 for CD. If you want a high res version, instead record at 88/24, process, resample to 44/24, then dither to 44/16 for CD. If you have no need for high res, don't bother with SRC and do it the first way, it's tradeoff is probably not better than the benefits of processing at 88 kHz.
www.digido.com and Bob Katz's book Mastering Audio will illuminate dither for you, I suggest you have a look there for more info.
brew,
Thanks a lot for the info. Yes, for quite some time I've recorded at 88.2/32bitFP, saved as 32bitFP, and later done all the processing in that domain. My editor can do the actual recording as a 32bit Float file. When you said "A better rule is to dither after any reduction in word length" it hit home- I know that's correct because I've read that many times before.
I had researched dither before and was unable to answer my questions, so after posting here I researched some more and was able to come to the same conclusion that you gave me. For the last several weeks I have been doing exactly what you described. My last steps have been:
> Back up a processed 88/32bitFP file to HDD as 88/24.
> Downsample the 88/24 file to 44/24.
> Apply a 16bit dither to the 44/24 file (if end product is to be a 16bit file).
> Split and save the 44/24 file to CD tracks. My editor can combine splitting into Redbook CD tracks and wordlength reduction as one operation.
You said:
"If you want a high res version, instead record at 88/24, process, resample to 44/24, then dither to 44/16 for CD."
The exact procedure of doing that and which dither to use can be confusing! I think you're saying apply a 16bit dither to the 44/24 file, correct? This is what had me stumped before. I need to set the L1 to 16bit dither and apply the 16bit dither to the 24bit file. To me this is counter-intuitive until you think about it awhile, then it makes sense.
Regarding hi-rez:
I really have no use for hi-rez. All I want to do is get my vinyl in the computer, perform noise reduction, and save as CD tracks. I find the idea of recording at 44/24 intriguing and avoiding having to re-sample seems to be a very good point, but I think this would mainly be to avoid overloading that can occur between samples on highly compressed and limited music, correct? Even so, it looks to me like the high sample rate isn't necessary at all, especially if I'm just going to throw half of it away at the end, right? All the hi-rez is doing for me is using up my HDD space and doubling the time it takes to process. The important thing is to have the 44.1kHz as a 24bit file. I do understand about division in calculations and how the word length increases. I think I'm going to start recording this way.
You said:
"I'd recommend setting the highest peaks for maximum of -2 or -3 dBFS to avoid distorting DAC's on playback".
Did you mean to say -0.2 or -0.3dbFS? In the Nielsen-Lund overload pdf you pointed me to, section 5 on page 12, regarding a fixed gain reduction as one solution to DAC overload they state "...a small headroom was found to be sufficient to avoid problems in the tested converters". The Waves L1 manual covers this in detail and they recommend -0.3db, so that's where I've always set the peaks at.
One (possibly) silly question:
I like to create mixes from assorted LPCD tracks by combining them into one long CD or a long file for later recording into my A810 reel to reel. I like to adjust the rms output of each track to be the same, as it makes for a very listenable composition. This is really the only reason I save the original hi-rez recordings. Thus, my question becomes "Is it possible to do a gain change to a CD track (i.e. 16bit track) with little or no signal degradation, what is the best way to do that, or do I need to start with the hi-rez backups to do that properly?" Many thanks for enlightening me!
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: