|
|
The following excerpts written by Jay Frigoletto first appeared in the rec.audio newsgroups on the internet over the past several years. They were in response to comments or questions asked about mastering related issues, and they are posted here essentially in their original form as they may be informative to those trying to learn about mastering.
The essays are brief, informal, and by no means a complete guide, but rather basic answers to questions posed by newsgroup participants. As they are conversational in nature and "off the top of the head", grammar and construction were not priorities.
1. What is mastering?
(This was a post to the rec.audio.pro newsgroup in late 1999)
Mastering is more than a fresh set of impartial ears, although this is certainly one advantage. It's the experience of the ears in mastering, which is different from mixing. The best mix engineers still get all of their work mastered. It's a different discipline, however related. It's the training of the ears, knowing what to look for, knowing how far to go, knowing how it will translate, knowing the monitors, the experience with so many records that have come through, with what works and what doesn't. It's the training, the knowledge that has been passed on by somebody more experienced, learning the aesthetic and technical details that aren't obvious or seem counter-intuitive to the lay person. It's the gear. Its the well designed and implemented monitoring environment, the high definition full range monitors, the knowledge of when and how to dither, what type, what gain, noise shaping or not, which curve. It's having the highest quality processing gear and the cleanest signal path. It's having digital gear that processes at a high enough resolution for the resolution of your project, and gear that doesn't truncate, and offers you dithering options when you need them, and has tried and tested quality algorithms and proper DSP practice. It's quality A/D and D/A and properly maintained and calibrated tape machines. It's understanding PQ codes, ISRC codes, noise reduction, and different master formats that will come in and be delivered. It's having the proper gear to make masters that glass can be cut directly from. It's knowing what glass mastering is in the first place, and why that's not done at the "mastering" houses we are talking about, It's knowing the difference between mastering and pre-mastering. It's knowing when to use analog or digital, how to get to and from digital with the least degradation. It's not getting carried away, it's restraint, but it's not necessarily timid. It's attention to detail, cleaning heads and tails, adjusting fades when necessary, making the album even from track to track, not thinking normalization will do this for you, understanding emphasis, not losing bits of data that are important or passing bits of data that are incorrect, knowing what to do about DC, phase relationships, balance, clicks, pops, dropouts, and how to prepare a proper log for the replication plant.
It's all of this and more. It is NOT simply putting an EQ and a limiter on the mix bus so your CD is loud and bright. If you don't understand the difference, you are sadly missing out on one of the most valuable assets in the completion of your project that you have put your heart and hard work into.
2. PQ editing
(This was a post to the rec.audio.pro newsgroup in early 2000)
PQ editing refers to editing the part of the subcode that contains info like track start IDs and end IDs, indexes, ISRC codes (stands for "International Standard Recording Code" and contains owner of the copyright, release #, record label...), emphasis, copy protection. It refers to the bits of the digital word (P & Q) that contain this info. You need to know about the red book spec and keep everything within it as far as the offset for the first track, minimum track length, maximum number of indexes or tracks per disc. Also, most PQ editors allow general offsets to help with universal playability - in other words, some old players take a few frames to unmute outputs when starting to play a new track, so you need to have the mark happen several frames ahead of first frame of audio. In the editor, you can place all the marks right on the first frame of audio, and the global offsets will make all your track start IDs the specified number of frames early. There are also splice and end offsets, and the track one offset has to be 10 according to the red book spec. Some programs like Jam or Toast automatically generate most of the stuff needed to have a disc play for you, but you can't really get in there and tweak. Other programs (or dedicated hardware from Sony in the old days) are standalone PQ editors that work in conjunction with your audio editor, and some things like Sonic Solutions have the PQ editor built into the same program (one reason Sonic is popular for mastering). You can place marks in the actual EDL (edit decision list - your audio editing window) and then the PQ editor will read those marks and generate a list, which you can then edit further like a regular standalone PQ editor, changing the times, offsets, ISRCs, emphasis, copy protection etc. Also, when making a master (1630, DDP, PMCD) the PQ info is put on the master somewhere separate from the audio so the plant can read it, check it against the PQ log you provide, and use it to cut the glass. Just because a CDR has the track start and stops on it when you play it doesn't mean you have a separate PQ file on the CD for the plant. The plant can get around that and extract it from the red book CD, but that's another step, and you know how that goes... Traditionally on a 1630, you would put a "PQ burst" (which sounded similar to a little modem tone just for a moment because it's basically a little data file) on the tape before starting audio (at 2:00 was pretty common). Now that 1630 is less common, it goes different places on DDP and PMCD, and you don't have to treat it as a little part of the audio. The programs usually just put it where it goes automatically.
3. ISRC codes
(This is a part of a post from 1999)
ISRC stands for International Standard Recording Code and is a 12 character code that
identifies tracks on CDs (and actually on music videos too). It identifies each individual track, not the whole album, so UPC catalog numbers can coexist with the ISRC.
An ISRC might look something like this: US REC 99 01360
And it identifies the following:
1. The country of origin (two characters, in this case, US, which is U.S.A.)
2. The Registrant (3 characters for the record company or producer/copyright owner)
3. Year (2 characters for the year that the ISRC was assigned, 1999 in this example)
4. Designation Code (5 characters assigned sequentially by the registrant). It will identify which song it is, maybe which album it’s from, or even which version.
The first owner of the copyrighted work assigns the code (provided they have already applied for and received a registrant’s code) which stays with the song for the rest of it’s life, whether the rights are transferred to another party or the song appears on a different album. The RIAA is the administrator of ISRC in the U.S.A. The code helps to identify a track, facilitating royalty collection, and can be used as a tool against piracy.
4. A clarification about noise shaping
(This appeared in rec.audio.high.end in late 1999)
[The following was the original comment from another poster]
> Noise shaping is NOT dithering. Noise shaping is a way to shape the
> quantization noise spectrum in a quantization process, whether or not
> the input signal that is being quantized has been dithered. Noise
> shaping and dithering are independent concepts and have different
> purposes.
[And Jay’s clarification]
...the difference in the original use of the term "noise shaping" vs. the more recent, and now quite common use of the term, which perhaps is better described as psychoacoustically optimized dither - or for short, noise shaped dither. The original use of the term (apart from the dry textbook definition: "a circuit which subtracts out the average noise value of a signal to increase the signal to noise ratio of the system") was usually in reference to oversampling. As you increase the rate of oversampling, you can spread the spectrum of noise (quantization error) out to higher and higher frequencies, so that much of the noise lies above the 20-20K audio band and is therefore inaudible. With 96Khz sample rate, you can similarly spread your dither out, thus having much of the noise above the audible range. And add to that the newer use of the term, and you can apply what is essentiual an EQ curve to the dither to distribute even more of it into areas where it is less audible (as per psychacoustic principles) and you can squeeze quite a lot of performance out of a system - more theoretical performance in fact than can be matched by current real world converter design. In the professional world this is nice because we can use a lower bit rate (say 18 or 20 instead of 24) while preserving as much resolution as current D/A converters can deliver, thus increasing available recording time on the media and lower data transfer requirements thus possibly alowing more channels or other information to also be transmitted (e.g. text, graphics etc).
5. Normalization caution
(This was posted to rec.audio.pro on Jan. 16, 2001)
[The following was the original comment from another poster]
> Normalization pulls the volume up such that the loudest single sample in
> your track is just below digital clipping. That's all it is--just a volume control.
> Everything is amplified equally including the noise.
[And Jay’s response]
As it's a non-trivial DSP operation, it also requires dither after the process to avoid truncation distortion. If this isn't accounted for in the software you are using, then you have a bit of a problem. Also, some normalization features alter the original file or create a new one, so that if you normalize and get a +3db change, and then do a volume change of -3db, you aren't back where you started, rather you've had two DSP operations, and possibly (very likely) without dither, so you've got a couple passes with truncation distortion and added noise. You've lost your original file and it's quality of sound forever. It can get ugly pretty quickly. By the same token, if you get +3db and then later decide it has to be a db less to match the other tracks better, but then maybe add 1/2 db because you got rid of a little too much, now you've had 3 DSP operations and your sound quality is going to be heading downhill very fast. The better programs (like Sonic Solutions for
instance) will always reference the original file and do the gain change in DSP in real time so no matter how many times you change the gain, it's applying the final changed value to the original file so that you only have a single DSP operation no matter how many times you change your mind. This still doesn't excuse you from needing to follow proper dither practice, however.
Another thing to keep in mind is that if you normalize and then perform further DSP, you have no headroom and will get clipping distortion. If you try to boost a db on the EQ, you don't have the db to spare, and you will clip. Even if you attenuate with the EQ, most filters will have a little ripple that will still cause some clipping distortion on the of the effected band(s). So the point is, don't normalize if you are going to do any further processing, and don't think you can just do a gain change and attenuate a db or two later if you decide you want to process the file unless you are sure that your software references the original file, or you're back to the problems discussed above.
> Normalization is usually a good thing.
Actually, normalization is a dangerous thing. Often, it can do more harm than good. If done properly as the last thing (other than dither) you do before having a CD replicated, it can be a good thing. Don't normalize something before sending it to mastering, and don't normalize without dithering, and don't normalize and then further change the gain. Normalization isn't quite as simple or cut and dried as it may seem. You still have to follow good practices and you need to know how your particular software actually deals with what it calls normalization. Hopefully this will help you avoid some of the potential hidden pitfalls.
6. Jitter in D/A conversion
(This was a post to rec.audio.pro in April, 2001)
[The following was the original question from another poster]
> Question is, especially regarding the Entech unit which is a highly
> recommended consumer unit with no wordclock i/o. If I've got my entire
> studio clocked properly via a good master clock source, should I worry
> about jitter in the last link in the chain before the monitor? Will a
> clock derived from SPDIF into the Entech unit for example introduce
> significant unwanted jitter.
[And Jay’s response]
I do think that all converters should allow external clocking, but they certainly don't, especially in the budget minded options. As for worrying about jitter at the last stage (the D/A), yes, absolutely. This is the place it will make the biggest difference to what you hear. You can have a room with some jitter in certain places and never have a real problem with the media you send out. In other words, you can produce good recordings even with some less than stellar jitter performance in some places because the gear on the other end can have a stable clock and resolve the stored info very well. But, to hear the best your room has to offer, an excellent D/A clock is important. The best scenario might be to use a D/A with a good internal clock and slave the rest of the room from that - except when recording initially when you would want to use the A/D's internal clock. If you have a high quality AD/DA box, problem solved. It's the same clock for both conversions. So, to try to sum up, if you are sending SPDIF to your Entech from a unit that is seeing a high quality clock, it will probably be just fine. If you had a D/A that accepted WC, using quality wordclock (from a distribution amp, not daisy chained) might make an improvement. But the best situation would be to have a D/A with a good clock as the master. One of the best things to remember about jitter is the often quoted statement from J.A. Moorer from Sonic Solutions which went something like this: Don't confuse the message with the messenger.