Warner DVD-A not full resolution (1 Viewer)

[Felix Martinez]

Sorry Felix but many of us can hear the improvement from 96 to 192khz. You may need a higher resolution system but you don't need an expensive one.

More power to you and other "golden ears." My own humble observation is based in the studio - ProTools, etc., etc. not necessarily home theater gear. I frankly can't afford some of the stuff I work with

My slightly tongue-in-cheek comment re: microphones was meant to imply one thing - which also relates to ultra high sampling, etc.; namely, there is *so* much that determines output quality other than sampling and word length, and these things are rarely the topic of discussion. Kind of like talking about how bitchin' an engine is, but the car shape is totally working against aerodynamics.

For someone to say "I hear a difference between X kHz and y kHz," more often than not - and I'm not addressing anyone here in particular - they mean between X and Y recordings. How in the world can someone make that claim listening to *two different recordings*?? Sometimes in different time periods, different equipment, etc.?? And when one *is* talking about the comparison between X recording on X format and X recording on Y format, unless the recordings were mastered simultaneously, where is the comparison? Even comparing the red book CD layer with the 2 channel DSD in hybrid SACDs is problematic as the red book is derived from the hi-res master and not mastered natively in 16 bit 44.1 kHz. I know some engineers that actually have two machines/computers, etc. capturing the same mix - one in high res, one in 16/44.1 to avoid conversions, etc.

The only vaild comparison IMHO is in the studio by actually recording a source in multiple formats using the same equipment and set-up. This is what I base my opinion on, and my opinion is mine only. Some will not agree, some might, but at least I know I'm not comparing apples & oranges.

Now let's listen to music!

Cheers,

 

[Felix Martinez]

Think of a 5 megapixel photo (DSD) versus a 1 megapixel photo (upsampling).

Isn't the analogy for DSD more like vector graphics than pixels? Then isn't it like lineart verses megapixels???

Just kidding!

Cheers,

 

[Brian Perry]

I'm afraid this is not a logical question-frequency response and sampling rate are two different events

Lee,

Could you clarify this? It is my understanding that the available frequency response of a digital recording/playback system is determined primarily--if not solely--by the sampling rate. (Nyquist theorem, etc.)

Also, how does "upsampling" differ from "oversampling"? It sounds to me as though they both mean interpolation, or educated guesses, of the missing data points between samples.

 

[ReggieW]

Guys,

We can go on about this indefinitely, but I honestly can't say if a recording suffers much with a lower sampling rate, especially if it was recorded in the studio at the sampling rate listed on the DVD-A. I have the REM "Automatic for the People" and it is probably the best sounding Rock DVD-A title I have in my library for stereo and multi-channel. Would it have sounded better at 24/96 instead of 24/48? I think that is a question only the engineer can answer, but "Automatic" sounds significantly better than its redbook counterpart to my ears. I too was concerned once I saw the low-sampling rate on the rear package that quality might suffer, but those doubts were erased once I got it home and gave it a spin.

Reg

 

[Lee Scoggins]

Would it have sounded better at 24/96 instead of 24/48?

Of course it would...think about this: you are recording twice the amount of musical information.

 

[Michael St. Clair]

re: dynamic range

I read a while back somewhere (didn't save a copy) that no direct live mic feeds even hit 90db.

Can anyone reference a direct live recording with a dynamic range greater than 90db? Even greater than 80db?

Of course, even if it is possible without artifical mixing (greatly mismatching levels of different tracks to create larger dynamics than were present in the original source...move noise floor of one source further from the peak of a different source) or expanding, the ability of higher res formats to 'capture' greater-than-Redbook dynamic ranges is something that will only affect classical and acoustic jazz. The pop/rock world is all about destroying, not expanding, dynamic range.

 

[Brian L]

Of course it would...think about this: you are recording twice the amount of musical information.

That assumes that a.) there is musical information on the source above 24K to begin with, and b.) that it is audible.

I really don't think that either of those two items have ever been proven, although they are of course subject to great debate, which we will not be resolving here

Using that logic, would a 24/96 recording of a phone conversation taken from the ear piece sound "better" than a 24/48 recording?

Methinks anything beyond 96K sampling is a waste of bandwidth. Arguments could also be made that anything beyond 48K is also a waste.

Brian

 

[Rich Malloy]

That assumes that a.) there is musical information on the source above 24K to begin with, and b.) that it is audible.

really don't think that either of those two items have ever been proven, although they are of course subject to great debate, which we will not be resolving here

OK, now I'm totally confused. Do the /48, /96, /192 increments refer to the pitch of the notes being recorded??? That's not right, is it??? I don't claim to know much about the technology of this, but I know music and musical acoustics, and while I understand that SACD is supposed to be capable of reproducing pitches above the CD limit of 20Hz, this nonetheless has absolutely ZERO to do with these various sampling frequencies.... right?!?

 

[Brian L]

OK, now I'm totally confused. Do the /48, /96, /192 increments refer to the pitch of the notes being recorded??? That's not right, is it

That would be the sampling rate, and theory holds (would that be Mr. Nyquist?) you will be able to record a frequency of 1/2 the sampling rate. So a 48K sampling rate will get you 24Khz, 96 will give you 48Khz, etc.

I am sure that John Kotches can explain that a whole lot better than I ever could. Or I could be totally wrong, and someone can correct me on that too.

BGL

 

[LanceJ]

That Nyquist limit works great.....for a pure sine wave at one frequency. But music is made up of multiple sine waves combined with one another--they are NOT all lined up neatly together. This means that an analog-to-digital convertor "looking" at a music signal 44,100 times a second (CD's sample rate) doesn't fully capture all those different waves. This is where even for me math/science doesn't explain everything (well, math/science not taken far enough).

From what I can tell this is why the CD format has problems with classical music: all those stringed instruments playing at once.

But with simpler music it can do quite well, as the new remasters reveal.

LJ

 

[Brian-W]

I swear if there was an audio review show like Ebert and Roeper has for films, Lee and John would be the ideal hosts.

Only problem is every show, they'd never get around to reviewing releases, just arguing endlessly about DSD vs. PCM.

 

[Ken_McAlinden]

Ken, I'm afraid this is not true. You get more resolution from lowering jitter and you also benefit from better mastering.

Jitter has nothing to do with resolution, though. One would hope that there are more cost effective ways of dealing with jitter than quadrupling the data rate. As far as mastering, up-sampling will give you slightly better results if you are mastering via digital eq and compression, but the results amount to losing less resolution rather than adding more.

Regards,

 

[Brian Perry]

OK, now I'm totally confused. Do the /48, /96, /192 increments refer to the pitch of the notes being recorded???

Rich,

Here's an analogy that may help:

The two main components of digital recording are the word length and the sampling rate. The word length (a.k.a, quantization or resolution) specifies the number of bits that will be used to measure the "value" of the analog acoustic wave. The sampling rate is the frequency at which the word length is sampled.

Let's say you wanted to measure the outside temperature over a 24-hour period. The word length would represent the precision with which you wanted to meaure it (degrees, tenths of degrees, etc.). The sampling rate would be how often you took the measurement (every hour, minute, etc.). Obviously, the more data you collect, the better idea you have of the weather. However, at some point, increasing the precision and/or the sampling rate doesn't really tell you anything significant. (I.e., do you need to know it was 82.45736278 degrees at 2:01:03:42 P.M.?)

Original redbook CD specified 16-bit resolution (word length) and 44,100 per second sampling rate for each channel. The reason 44,100 was chosen by Sony and Philips was because theoretically you need two samples to capture a given frequency. As human hearing doesn't go much past 20,000 Hz, the thinking was that 22,050 (44,100 div. by 2) was a safe margin. Likewise, 16 bits allows for up to 96 dB dynamic range (more with proper dithering, but that's another issue...)

As electronic technology has improved, the high-res formats have raised the top specs. For DVD-A, the benchmark is now 24-bits resolution and 192,000 samples per second. Thus, a 192k sampling rate could capture a frequency of 96,000 kHz. But is there a reason to? The question is whether these increases make an audible difference. And, as others have mentioned, are recording engineers actually needing less bandwidth or data due to the trend of compressing the dynamic range of recordings to make them sound "louder"?

 

[Rich Malloy]

Excellent explanation--I think I sorta understand in a somewhat oversimplified way!

I'm still fuzzy on the word length/bit rate part. Between 16 and 24, say, what additional "value" is captured by the longer word length/higher bit rate?

 

[Brian Perry]

Between 16 and 24, say, what additional "value" is captured by the longer word length/higher bit rate?

If you convert binary to base-10, 16-bits translates to 65,536 possible values for the signal. In other words, the value could be between 0 (16 zeroes, "full silence") and 65,536 (16 ones, "max volume"). This resolution is capable of 96dB of dynamic range (the difference between the loudest and quietest sounds), which was thought to be more than we'd ever need. In contrast, your local phone calls are the equivalent of 8-bits.

Each time you increase the bits (8 to 9, 15 to 16, etc.) you double the resolution. Therefore, if you go from 16 to 24 bits, you're going from 65,536 possible values to over 16 million. While impressive, the issue is whether the human ear can differentiate between such minute gradations. And also whether the potential dynamic range (144dB for 24 bits) is beyond the thermal distortion limits of just about all amplifiers and other playback equipment.

(I should mention that there are advantages to using higher numbers, in much the same way a car that can go 150 mph will be smoother at 55 mph than a car that has a top speed of 55. For example, many of today's players built for 24 bits can decode 16-bit data with fewer errors than older CD players.)

 

[Lee Scoggins]

And, as others have mentioned, are recording engineers actually needing less bandwidth or data due to the trend of compressing the dynamic range of recordings to make them sound "louder"?

If we confine ourselves to an "MP3 world", we don't need any of this stuff. But if we care deeply about music, this technology is as important as the PCs chip revolution.

 

[Brian Perry]

If we confine ourselves to an "MP3 world", we don't need any of this stuff. But if we care deeply about music, this technology is as important as the PCs chip revolution.

Agreed.

 

[Al B. C]

My head hurts.

What's that old saying.......

"Fuck art, let's dance!"

 

[Rich Malloy]

 

[Al B. C]