What is transistions?

W

WarmJetGuitar

New member
Hey there... it's been a while... hope you're all doing well out there.

I was reading this article where Mike Spitz mentions a term called "transitions" as a way of measuring audio quality. It left me wondering what that means? Havn't been able to find anything usefull on Google, though I might not be searching well enough. Are any of you familiar with this term and able to explain it to me?

The article wasn't from a recording specific website though Mike Spitz was if any a credible source.

"Analog’s attraction lies in its ultra-high resolution capability, Spitz explains. Direct Stream Digital (DSD), the high-resolution digital disc format Sony used for its audiophile SACD format, is capable of 2.884,000 transitions per track per second, but a high-quality mastering tape contains approximately 80 million transitions per track second. “And that’s just for 1/4-inch two-track tape running at 15 IPS,” says Spitz. “The resolution goes up substantially with wider tracks and higher (tape) speeds.”"
 
Well, never heard of 'transitions' in this context, but it's maybe pretty obvious what is meant.

As a comparison, I wonder how many transitions the human ear/brain system is capable of.

I would suggest that one transition is a smallest possible change in anything connected with a sound, whether it be frequency, volume, whatever. So am absolutely steady single note would have two T, namely start and stop. if the note has a more complex envelope, then there could be more T for the attack, and maybe many more for the decay. A more complex change could have many more T. Multiple notes, multiple instruments, echo, ambiance, etc, etc, will all add many more T, hence getting up to the sort of figures mentioned in the quote. My guess is that the ear/brain can detect many more T that it might be fully aware of, such that one might be aware that there's something 'not quite right' with a sound/piece of music even if you cannot say quite what's wrong?

A midi file in effect stores the mechanics of such transitions, although many T might be 'fixed' into the sound produced by the sound generator rather than being fully under the control of midi. Think how many midi events there might be in a midi file! And that's NOT counting anything to do with ASDR, or effects, or vibrato, or anything else that's controlled 'en masse' by the sound generator.

Spitz is stating that magnetic tape (dep on tape width and speed) is capable of defining many more T than any digital process. Could be so. I would guess that the ear/brain capability must lie somewhere between the digital limit, and the tape limit, otherwise digital recording would never be acceptable to the human ear.

Geoff
 
DSD uses a 1-bit PWM-like system. Assuming a 5.6MHz sample rate (DSD128) and Nyquist's theorem, that would allow it to sample 2.8 million on/off cycles a second.
 
Thanks JP :-) Wouldnt claim that I truly got my head around it though it kindda makes sense. It's curious that there's not a commonly used term to describe this way of measuring audio quality.
Could be one of the reasons why many people like the sound of tape better? As it seems digital systems have caught up with even the best tape recorders (on least on standard formats) in terms of frequency response.
 
The only problem with the notion that having more "transitions" leads to be there sound quality is the fact that we are sampling audio. 2.8 million is far more than necessary, since humans can't hear much over 20,000 cycles. So, having over 200 times the resolution necessary to store anything in the audio range, and significantly beyond, is going to be just fine.

Having 40 times more resolution than that will not add anything

This is the typical horseshit that people believe in order to justify their feelings. The feeling is valid, the justification isnt.

Just because someone has a history of making great recordings, doesn't mean that they have any idea how things like this work. The same way that most of us successfully drive a car to work everyday, but could not explain how an automatic transmission works.
 
DSD is a different way of sampling. It doesn't use 16 bit, 24 or 32bit word length, it's only 1-bit word length. I think of it as eternal dithering.... lol.

Spitz is stating that magnetic tape (dep on tape width and speed) is capable of defining many more T than any digital process. Could be so. I would guess that the ear/brain capability must lie somewhere between the digital limit, and the tape limit, otherwise digital recording would never be acceptable to the human ear.
Click to expand...

I think it is a bit of a misnomer to say tape has 80 million "transitions" per second. It's not like your altering each magnetic domain individually to represent information. There might be 80M magnetic domains, but they are all representing the same info, just enmasse so the signal can be detected during playback.

And it's not like a digital reproduction is playing back the discreet samples and your ears smooth it out. The output amps have smoothing filters; you're hearing a true analog signal when you listen to a digital source. Video is one frame at a time and your eyes have persistence which makes the video look smooth, but digital sound doesn't work the same way.
 
There is nothing jagged about digital audio that needs to be smoothed. The filters are the nyquist filter that gets rid of everything above half the sample rate. If the highest audio frequency you are trying to record is below the nyquist frequency, no more resolution is needed to reproduce it.
 
WarmJetGuitar said:
Thanks JP :-) Wouldnt claim that I truly got my head around it though it kindda makes sense. It's curious that there's not a commonly used term to describe this way of measuring audio quality.
Could be one of the reasons why many people like the sound of tape better? As it seems digital systems have caught up with even the best tape recorders (on least on standard formats) in terms of frequency response.
Click to expand...

Thinking about it, it's not quite such an apples-to-oranges comparison as it first seems.

DSD uses pulse-density modulation rather than usual PCM, hence it records a single bit at a stupid frequency and somehow stores the level in the time domain rather than directly recording the level as a number.

The result is something like this:

Pulse-density_modulation_2_periods.gif
(source: https://en.wikipedia.org/wiki/Pulse...a/File:Pulse-density_modulation_2_periods.gif )

Now. If we take a piece of analogue tape and coat it with magnetic developer, we end up with something eerily similar:
tapetrk1.jpg
(source: http://www.audio-restoration.com/tapetrk1.jpg )
 
You must log in or register to reply here.
Back
Top