Home |
Search |
Today's Posts |
#1
Posted to comp.os.linux.advocacy,rec.audio.pro,alt.audio.pro.live-sound,comp.os.linux.misc
|
|||
|
|||
Analog vs Digital?
Mark Kent wrote: begin oe_protect.scr chrisv espoused: Mark Kent wrote: chrisv espoused: Of course there is. If the output of a record/playback system matches the input, then it is "accurate". "How accurate" the system is is easily measured in terms of added distortion and noise. No, it's far from easily measured, it's extremely difficult to measure. Indeed, huge amounts of work have been done on digital codecs and their impact on sound, much of it by my good self over the years, using a technique known as 'subjective testing' where hundreds of people are wheeled into specialist soundproof rooms, given samples to listen to, or channels to speak over. They then vote, and a MOS chart is produced. Using latin squares to randomise the samples against people produces at completely predicatable curve shape each time, the interesting bit being where the curve sits. Even just measuring 'distortion' alone is incredible difficult - have you ever tried to use an analogue distortion meter? Unless the metering devices are signficantly better than the devices they're measuring, the results are meaningless anyway. If you're dealing with top-quality equipment, there's only one final test - your own ears - nothing else is remotely good enough. But you need to know what you're listening for. It appears that you are talking about the field of codec development, which is not what I've been talking about. Of course 'subjective testing' is needed in that field, to determine how much loss you can get-away with. No, I'm talking about measuring quality, which you don't seem to know much about. The use of subjective testing is because it's virtually impossible to /measure/ these things. I've been talking about digital as a method of maximum-quality recording, and it's ability to do this very well (and, as a corollary, CD's quite-good ability to serve as a music-distribution system). There is no such thing as 'maximum quality'. Certainly, when you down-sample to 44kHz so that you can produce the retail CD, you're have to lose the frequencies above 22kHz (there's really nothing there anyway - microphones don't even go that high), but that's not a problem with digital audio in general. Yes Microphones do! They are not /digital/ in operation, they are highly analogue, just as music is analogue, not digital. They do not suddenly stop working at a given frequency, they have a gentle roll-off. You'll also find that good loudspeakers work well above this frequency, too. I didn't mean to imply that the "suddenly" stop working. However they are well-past their upper-bandwidth point, as defined by half-power, by 22kHz. But this is a rather critical point, isn't it? The -3dB point is just that, a shedload of sound is being filtered out here. I'll conceed that bandwidth limit at 22kHz is a minor drawback with the CD format (but not with digital audio in general). It's a major drawback. But really, there ain't much up there, and even if there was you probably couldn't hear it anyway 8). There's a lot up there, in fact. But you now seem to be arguing that certain types of distortion in digital recording don't matter... The limits to the number of A/D/A/D/A/D/A/D/A/D conversions you can do even using the /same/ codec are determined by two key factors, assuming reasonable quality engineering all round: 1) Cumulative quantisation distortion 2) Cumulative clock-noise I think that we can agree that running a signal through multiple imperfect processes can degrade the signal. Of course. Good... In particular, you cut the bit about quantisation distortion, which suggests to me that you don't understand it. No, I have addressed that non-existant problem. 8) It's a real issue. So trivially solved that it cannot be considered a "drawback of digital audio". The quantization errors are swamped by noise, either already-present or added as dither, resulting in nothing but random or psuedo-random noise. Noise that even on CD's 16 bits is can be 93dB down from peak-level i.e. not audible. (The noise you do hear on some CD's is invariably from a analog master tape.) It's not solved, it's masked by the addition of what is essentially a noise component, selected to make it pleasing to the ear. Many people have asked why there are only two controls on my hifi (volume and input). The answer is - why would I want any others? I can't live without a bass control - there's too much variability among recordings. But that's another argument and I think we've argued enough. 8) The other thing which you've not mentioned, but is very important, is the room you play things back in. The acoustics of that room are also acting as a filter on the original sounds, too. There are many thing we've not covered, obviously. It should not be assumed that we are ignorant of the issues not covered. I think I've made my point enough times now... -- | Mark Kent -- mark at ellandroad dot demon dot co dot uk | My only love sprung from my only hate! Too early seen unknown, and known too late! -- William Shakespeare, "Romeo and Juliet" |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
on topic: we need a rec.audio.pro.ot newsgroup! | Pro Audio | |||
Artists cut out the record biz | Pro Audio |