Does analog move more air. . . ?

[Tim Gillett]

Okay, but what about the fact that A/D & D/A converters sample the Analog waveform at finite intervals, and at high frequencies there is a compromise?

What about this quote:

"Once set, sample rate does not vary during a recording, although different audio files recorded at different sample rates may be used together in a multitrack system if the software permits it. Usually, as in the case of a DAW, audio files of differing sample rates will need to conform (be converted too) a single sample rate, typically 44.1KHz, 48KHz, 96KHz or 192KHz. This sample rate is usually set in the application preferences for the recording session.
■Higher sample rates produce better quality recordings but also bigger file sizes which demand greater space on storage devices (such as hard drives), and faster processors (CPUs) to manipulate.
■Lower sample rates produce poorer quality but also smaller file sizes which demand less of storage systems, CPUs and will transfer over networks (internet) faster."

VP

Yes it samples at finite intervals, but finely enough to detect perfectly the highest frequency that you want to record. The intervals finer than that are for higher frequencies than you have chosen to record. If you want to record them too, choose a higher sample rate. That's all!

The quote says "higher sample rates produce better quality recordings". That's only true up to the limits of practicality. The phone networks only reproduce audio up to about 3khz, or used to. The upper limit for full frequency audio is generally agreed as 20khz. Do you want to record frequencies up to 40khz? Use 88khz sample rate.

But again, no matter what sample rate you use, you'll never record a true square wave.

Tim

 

[grimtraveller]

At this point, I'm totally lost. Whatever focus I might have had a couple of pages ago has departed. I can't recall which what whom is trying to prove.
All I can gather is that Pete thinks digital is the devil {after all, they both start and finish with the same letters ! }, A to D regrets starting this thread and ironically, in the end maybe Lonewhitefly is right; maybe no one who has sullied their hands with the dreaded D word should ever come into the analog only forum, given that anyone that has partaken of "D" clearly is no longer bursting with analog purity.

 

[mshilarious]

Oh I am bursting with analog purity! But analog to me <> tape. There are all kinds of phenomenon that may exist in pure analog electronics that are tough to deal with in any recording medium, although ultimately digital can get closer. With digital it's easy to do things like cheat to get enormously large dynamic range and frequency response.

For example, Neumann in their digital microphones uses two channels of ADC, one loud and one soft, and then constructs an output signal from the two. That's pretty easy DSP; when loud signal is > say half-scale (or whatever), then you switch to soft signal digitally amplified (which is simple multiplication). They output a 150dB dynamic range, which is pretty close to the maximum you'll get out of any analog circuit. It's more than the microphone can handle anyway. Do that at 192kHz and now you have 150dB dynamic range across 80kHz bandwidth!

 

[mshilarious]

Higher sample rates will sample the waveform more times per cycle, resulting in a more accurate "rendition" of the recorded signal. Of course that would make a more accurate output signal, that is a fact. That quote was from a credible website and has a lot of merit despite your "Naysaying"

Only if you define accuracy as greater bandwidth, which you can if you like, then you just have to use a higher sample rate. There is no increase in accuracy beyond a very small transition band attenuation within the smaller bandwidth of the lower sample rate.

That is to say that you won't get any better accuracy at 18kHz going from 44.1kHz to 88.2kHz, but you'll get far better accuracy at 36kHz, going from none (44.1kHz) to flat response (88.2kHz).

Then you just have to decide how much bandwidth you need, and act accordingly.

You have already proven to yourself that digital is perfectly happy producing sine waves at 10kHz and 44.1kHz sample rate, which I don't think you believed at the beginning of this thread. Try a 19kHz sine and see what you get.

 

[Victory Pete]

Yes it samples at finite intervals, but finely enough to detect perfectly the highest frequency that you want to record. The intervals finer than that are for higher frequencies than you have chosen to record. If you want to record them too, choose a higher sample rate. That's all!

The quote says "higher sample rates produce better quality recordings". That's only true up to the limits of practicality. The phone networks only reproduce audio up to about 3khz, or used to. The upper limit for full frequency audio is generally agreed as 20khz. Do you want to record frequencies up to 40khz? Use 88khz sample rate.

But again, no matter what sample rate you use, you'll never record a true square wave.

Tim

Okay, thanks for your helpful responses. I understand about the square wave. I am still convinced Digital is "Embellishing" high frequencies though.

VP

 

[Victory Pete]

Only if you define accuracy as greater bandwidth, which you can if you like, then you just have to use a higher sample rate. There is no increase in accuracy beyond a very small transition band attenuation within the smaller bandwidth of the lower sample rate.

That is to say that you won't get any better accuracy at 18kHz going from 44.1kHz to 88.2kHz, but you'll get far better accuracy at 36kHz, going from none (44.1kHz) to flat response (88.2kHz).

Then you just have to decide how much bandwidth you need, and act accordingly.

You have already proven to yourself that digital is perfectly happy producing sine waves at 10kHz and 44.1kHz sample rate, which I don't think you believed at the beginning of this thread. Try a 19kHz sine and see what you get.

Well I am not sure those sine waves produced are not "Fabricated"

VP

 

[Bobbsy]

PS An over driven guitar signal comes close to being a square wave, I can use that as a test signal also.

There is a world of difference between "close to being a square wave" and a true square wave with it's theoretically infinite frequency on the rising signal.

Try your experiment with the over driven guitar.

 

[Tim Gillett]

Well I am not sure those sine waves produced are not "Fabricated"

VP

"Fabricated" or not, they are an accurate reproduction of the input. You cant ask for more than that. In a way I dont care HOW it works, it so long as it works. And it does work. As John Wayne used to say,"It's good enough for me."

Tim

 

[Victory Pete]

There is a world of difference between "close to being a square wave" and a true square wave with it's theoretically infinite frequency on the rising signal.

Try your experiment with the over driven guitar.

Okay yes I understand about a true square wave. I will take pictures of the overdriven guitar signal.
VP

 

[Victory Pete]

"Fabricated" or not, they are an accurate reproduction of the input. You cant ask for more than that. In a way I dont care HOW it works, it so long as it works. And it does work. As John Wayne used to say,"It's good enough for me."

Tim

So you dont find Cymbals "Harsh and Brittle"?

VP

 

[Bobbsy]

I find cymbals harsh and brittle when I listen to them live (my son used to be a drummer in his teen years) and a good digital recording reproduces this fairly accurately.

Many analogue recorders are rolled off and slightly muddy on cymbals which, in some cases can be a benefit and improve the overall sound.

 

[mshilarious]

Good cymbals recorded with flat-response microphones are not harsh. Cheap cymbals recorded with microphones that have large HF peaks and troughs are probably not going to sound good at all or at least without a serious amount of equalization.

 

[AtoDeficient]

I seem to remember hearing that all those spectacular vivid color pictures that the Hubble Space Telescope took were actually, at best, originally black and white, and were quite probably mathematically generated and computer-enhanced, in that they weren't actually "seen" by the telescope visually, but created with bits of digital information the telescope received. . .

I just thought I'd throw that in here. . . I figured, what the hell. . . This thread has lost all its focus anyway. . .

 

[Victory Pete]

Good cymbals recorded with flat-response microphones are not harsh. Cheap cymbals recorded with microphones that have large HF peaks and troughs are probably not going to sound good at all or at least without a serious amount of equalization.

This what I have on my house drum kit Zildjian Armand Zildjian Series 4 Cymbal Set-Up | Sweetwater.com I record them with these Shure SM81 | Sweetwater.com they dont sound "Harsh and Brittle" until I mix down to DAT or CD.

VP

 

[mshilarious]

Those are made either by compiling individual images taken through different color filters, or by using the visible spectrum to represent invisible frequencies.

If you are interested in amateur astrophotography, digital technology has created a major revolution in imaging with small telescopes, much more profound than the stuff we are arguing about with audio recording. I mean you can put up a high-quality tape vs. digital recording and there is at minimum an argument. But now with modern software that can combine literally hundreds of CCD images into a single image vs. what you could do with a hand-guided long-period single film exposure is not even a question. You'd have to be blind to not see the difference:

Starizona's Guide to CCD Imaging

 

[mshilarious]

Exactly how old are your DAT recorders? It is entirely possible that the filters on a DAT machine are not up to modern standards of quality.

 

[Victory Pete]

I seem to remember hearing that all those spectacular vivid color pictures that the Hubble Space Telescope took were actually, at best, originally black and white, and were quite probably mathematically generated and computer-enhanced, in that they weren't actually "seen" by the telescope visually, but created with bits of digital information the telescope received. . .

I just thought I'd throw that in here. . . I figured, what the hell. . . This thread has lost all its focus anyway. . .

"Lost Focus" that is funny! I am an "Amateur Astronomer" from when they launched the Hubble, I bought a Meade 10" LX200
http://www.bhphotovideo.com/bnh/controller/home?O=&sku=616997&Q=&is=REG&A=details Digital is great for Astronomy, the CCD imagers that have evolved in the last 20 years are amazing and they are definitely Digital, Digital definitely has its place. Obviously without Digital we could not have music on the internet. What I have a problem with is all the "Hype" about Digital this and Digital that, while Analog was sent out with the trash, that is the irony. Pretty soon they will be marketing "Digital Toilet Paper" It is digital though: 1 1 1 1 1 1 1 1 1 until the roll is empty: 0 0 0 0 0 0

VP

 

[Victory Pete]

Exactly how old are your DAT recorders? It is entirely possible that the filters on a DAT machine are not up to modern standards of quality.

Well when was the last time you saw a DAT machine being produced? Mine is the Tascam DA-30 MKIII. My CD recorder is only 3 years old. What exactly is the "Modern standards of quality"? MP3's and the like?

VP

 

[mshilarious]

No not .mp3, don't be silly obviously those are horrible at the bitrates you get from the download/streaming services.

At least three things have changed since the digital stuff of the '90s: first, converters have improved to enjoy more dynamic range. 100dB or less was typical 15 years ago; today the best converters are about 120dB, and prosumer converters commonly exceed 110dB.

Second, converters run at higher speeds--the true sample rate, not the output data rate--this means more oversampling which can yield a higher quality transition band performance.

Third, DSP has gotten cheaper as the semiconductor industry has helpfully crammed a *lot* more transistors into an IC. Literally hundreds of times more. See, oversampling doesn't do a lot of good if you can't process that data in close to realtime to generate an output rate. Decimation filters are an approximation of the very good math that generates the excellent transition band performance, yielding really low aliasing and very low passband attenuation.

It's like my square wave Excel sheet, it's a limit equation I am attempting to solve to create that square wave. The more harmonics I add the better it gets. If I pump the thing up to >100 harmonics from the 14 I did, and if I increase the resolution from 5 degrees to 0.1 degrees then I indeed get something that looks a whole lot more like a square wave. That makes a spreadsheet so large I'd need to set manual recalculation on my old 1991-era 386 (long gone), but no worries for my also long-in-the-tooth P4.

Similar idea with a decimation filter; give it more input (oversampling) and a lot more processing power and it's going to do a better job.

In my experience the point of diminishing returns was reached around 2003; many converters based on older chips have measureable limitations that *might* be audible, whereas stuff based on later pro-grade chips is usually pretty good. Digital is getting to be mature as there isn't much more that can be done within the limits of physics, certainly with respect to audibility. When your aliasing products are already below -130dBFS there is only so much more to do . . .

 

[Victory Pete]

Well then it is time to switch to Analog, real Analog not "Toy Analog"!

VP