Today's lecture / presentation left me with a lot of questions that I decided to reserve for this post because I was afraid I'd end up taking up too much time.
Oh yeah, and by the way, terribly sorry about being late to class.
Image formats
Something I've run into that I want to know more about is JPEG 2000. It's not the same as the old JPEG standard. I think, if memory serves, that JPEG 2000 is used by the Pitt ULS for storing preservation masters instead of TIFF because it's capable of lossless compression. (I think.)
Something else: GIF handles simple animations; PNG does not. I don't know how relevant that is to the course, but....
Sound formats
First of all, on audio quality, I thought that CD-quality audio is 44.1 kHz, not 22 kHz. Then again, the Wikipedia article I linked says something about a 22.05 kHz Nyquist frequency and I have no idea what they're talking about. If someone could help me with this I'd appreciate it. I know it isn't terribly relevant to the class, but it's going to bug me.
Second, the only audio format we talked about was the infamous MP3. But it seems to me that MP3 is on its way out - largely because of its lack of support for Digital Rights Management, but also because better formats are now available. AAC files are smaller and better than MP3. There's also Microsoft's WMA, RealAudio, and the lesser-known Ogg Vorbis and FLAC.
Identifying digital objects
I'm really unclear on this - PURL and DOI sound more like address forwarding than identifiers. From the lecture it sounded like those services would just make sure that people could always find the items at their original location. (Valuable, but not all that an identifier does.) In my online travels I've seen use of MD5 checksums used as an identifier of sorts - if they match, it's the same item. I don't know - maybe I'm overthinking it.
Assignment 2
Just one minor thing here: do we have to use flickr? If we already have accounts with another image-sharing service, like Google Picasa (which conveniently integrates with Google Blogger) would that be acceptable too?
Muddiest point week14
16 years ago
4 comments:
CDs are sampled at 44.1kHz, yes, so that means the highest frequency represented is, roughly, 22.05kHz. Which is just a smidge above what people can really hear.
Nyquist was this Information Theory* guy, who determined that to perfectly represent a sound you need to sample it at twice the rate of the highest frequency represented--what we now call the "Nyquist frequency." Higher is fine, but it gains you almost nothing. Lower, and you begin losing sound quality quickly, as frequencies are cut out. I'm not sure I can make this make sense in any intuitive way without drawing pictures--or possibly even then--so it would be easier if you could just accept it as fact. :) If you can't, I'll see what I can dig up.
A bit of a digression: Sampling is an Easy Problem. Quantization is a Hard Problem. The sound quality that is lost, in converting from analog to digital, is lost in quantization. Basically, you can't represent every analog voltage level (because, let's be honest, when you put sound on a wire, into a speaker or headphone, it's voltage) precisely, so you have to pick discrete levels. There are a lot of them in the 16-bit (I think) CD representation, so you don't lose much. But it's comparable--if you'd like to think of it that way--to GIF's color/palette-picking problems: since you have to pick from a set of values, you do lose some information. The particular way CDs do it favors quiet sounds over loud ones, slightly. (This is why those audio crazies claim that records have better fidelity than CDs, by the way. They are wrong, since records are so easy to damage, and the quantization errors are so small, but it makes them happy, so whatever.)
*Very different than Information Science. More mathy. Electrical/communication engineers and some high-power computer scientists (who are, themselves, really communications people) study it, which is why I know it at all.
Coral, thank you for explaining that to me concisely (and so well).
I'm curious, though - one of the things that Beatlemaniacs (like myself) are waiting for are remastered versions of the original albums. Thus far we've gotten a number of tracks remastered (Yellow Submarine and three of the Beatles movies) and some remixes (Let it Be...Naked, Love) but the rest of the catalog is reportedly still being worked on.
(You probably knew this, but the original albums were quite compromised even before conversion to digital because audio technicians only had four tracks to work with. So everything had to be "mixed down" and fidelity was lost. I would imagine, based on what you've said, that the mixing process could've done some damage - given that different singing voices / instruments operate at different frequencies and mixing them together would mean that things might get chopped off in unfortunate ways.
Now things are quite different - it seems they have an unlimited number of tracks to work with, so each track could have its own frequency levels (if I'm understanding this correctly).
So what I'm asking (in a long and boring way) is: is this why remastering can make such a great difference?
We're beginning to leave the land of my expertise, but I can at least make a few comments and guess at things...
I suspect that audio engineers have tools available to them that can "fill in" some lost values and improve the overall sound, not entirely unlike the way in which remastered video is made to look so nice, despite damage. Or, to be more relevant, similar to how we can figure out what was erased from documents after they are digitized, with processing software.
If there were only four tracks to begin with, that's really all there are now, though there may, of course, be any number of copies of them, all filtered and being processed in interesting ways.
The combination of frequencies and how that all gets mixed together and digitized is one of those things I wonder a lot about, myself. It seems to me that, if you have a bunch of instruments playing together, you'll get what are called "beat frequencies," and I wonder whether those might not be lost in the album creation process, nowadays. I'm honestly not sure.
Coral,
I may be mistaken, but I believe that the Beatles / Apple Records kept the original master versions of each track. I'd have to look into it, but I don't have much time and I wouldn't know where to even look.
Post a Comment