The No.1 Website for Pro Audio
20th January 2020
#601
Gear Guru

Quote:
Originally Posted by Tom Barnaby
When the system takes a sample of the amplitude of the global signal, it measures the amplitude of the addition of many signals.
Right. When they're added it's ONE signal. ONE waveform.

Quote:
Originally Posted by Tom Barnaby
The interesting point is that two signals don't behave like one if you consider them is the time domain, because both waves did not start at the same time.
This is the reason why we have to study very carefully what happens between two sampling points to see if some audible information can be lost
No, you're completely missing the point. The analog electric signal moves between two points over time. TWO. NOT many. It's a signal that alternates between two directions. Not three. Not four. Not fifteen.

You're thinking of "time domain" versus "frequency domain" completely incorrectly.

If you were right about this then in order to capture what you're talking about not only would the microphone's diaphragm have to move accordingly, and not only would the coil in the magnetic field have to transduce that, but the electrical signal running through the cable would have to represent that with more than one amplitude over time....

.... yet that's not what's happening...

Quote:
Originally Posted by Tom Barnaby
The image in your post above represents what happens with one signal and its harmonics, but this is not real life audio.
You're completely missing the point. Any event that happens and "concludes" between samples is either a simple sine wave or a complex waveform which in turn consists of simple sine waves. Regardless, if they are concluded between samples then they are by definition higher frequencies than we're sampling. That's it. There's nothing else to that. They're just higher frequencies.

When you're looking at the image I posted do not think of that as any frequency, think of it as the highest frequency we want to capture. Both it and any lower frequency can be accurately captured and described, with a pretty great resolution of phase.

There is only one amplitude at the point we're taking a sample. At any point at or between samples. Not many amplitudes. Just one.

When the waveform becomes complex and no longer is a sine wave then it - and all the component sine waves that make up the complex one - are below half the sampling frequency and the resulting complex waveform that you see - while jagged - will be accurately represented for the aforementioned reasons.
20th January 2020
#602
Lives for gear
Quote:
Originally Posted by Tom Barnaby
96 khz sampling doesn't make you hear frequencies that are not audible, but is enables a more accurate reproduction of the analogue signal.
But we didn't have an accurate reproduction of the analogue signal until digital came along! That's what most people hate about digital - it tells the truth unlike the various historic approximations of waveforms ingeniously stored on various physical media - each stage of production towards the finished media added distortion, saturation and rolled a little more top-end off. This happened to be pleasing to the human ear - it certainly isn't accurate. Making digital even more accurate won't necessarily make it sound "better" to the punter on the street - scientifically better perhaps - but musically?
20th January 2020
#603
Gear Maniac

Quote:
Originally Posted by Yannick
So your theory is that a wave has some sort of memory, some identity ?
Maybe it is an intelligent lifeform ?

Study the article and video I posted above, please, before you post anymore
I don't think you understand what I mean.
The question with sampling should always be what does happen between two sampling points.
It's a fact the signal always exists between sampling points. What we need to know is if this information is useful or not to reproduce the analogue signal properly after the sampling process.

You seem to think that this information is not useful as long as we follow the Nyquist principle. I think that the problem is more complex, as Nyquist worked about frequencies and did not really care about their distribution in the time domain.
For telecom applications, the Nyquist principle is relevant, but for music it can be a limitation.
20th January 2020
#604
Gear Guru

Quote:
Originally Posted by Tom Barnaby
I understand what you mean.

It's true that there is only one physical waveform, but we are here talking about the complexity of this waveform. A waveform resulting of the addition of two primary waveforms is more complex than a single waveform.
We have to consider this complexity not only about the frequency content, but also, and it's more important, in terms of evolution of the amplitude in the time domain.
Sound is changes in air pressure over time, picked up by our ears where components move back and forth over time.

Analog audio in a cable is changes in voltage over time.

Your "amplitude in the time domain" and "frequency content" is the same thing. Frequency is X over time.

The "complexity" can be thought of as the shape of the waveform. A square wave is more complex than a sine wave. Yet they're both made up of amplitude changing over time.
20th January 2020
#605
Gear Maniac

Quote:
Originally Posted by mattiasnyc
Sound is changes in air pressure over time, picked up by our ears where components move back and forth over time.

Analog audio in a cable is changes in voltage over time.

Your "amplitude in the time domain" and "frequency content" is the same thing. Frequency is X over time.

The "complexity" can be thought of as the shape of the waveform. A square wave is more complex than a sine wave. Yet they're both made up of amplitude changing over time.
You answer is very useful because it will help us to tackle a key question.

You say that the amplitude in the time domain and the frequency content are the same thing. This may look trivial, but I think it is not.

amplitude variation and frequency content are not the same. For example, the addition of two sine waves of same frequency is another sine wave of same frequency only if both waves are perfectly aligned in the time domain ?

What does happen if both sine waves are not aligned ?
What is the form of the resulting signal ?
20th January 2020
#606
Gear Guru

Quote:
Originally Posted by Tom Barnaby
You answer is very useful because it will help us to tackle a key question.

You say that the amplitude in the time domain and the frequency content are the same thing. This may look trivial, but I think it is not.
Well you're wrong. "Frequency" to us means nothing if it has zero amplitude. And there is no frequency if there aren't changes over time. An audio frequency is a signal whose amplitude varies over time. It's that simple.

Quote:
Originally Posted by Tom Barnaby
amplitude variation and frequency content are not the same.
Sure it is.

I think you're maybe looking at data on a computer and having it shown in the frequency domain versus the time domain but that's not the way things really are in real life though. Those are views of data (re)constructed into those separate views. Not "reality".

Quote:
Originally Posted by Tom Barnaby
For example, the addition of two sine waves of same frequency is another sine wave of same frequency only if bothe waves are perfectly aligned in the time domain ?

What does happen if both sine waves are not aligned ?
What is the form of the resulting signal ?
It doesn't matter what the form of the resulting signal is. It's still just ONE amplitude varying over time. The fact that it doesn't look like a sine wave is irrelevant.
20th January 2020
#607
Gear Maniac

Quote:
It doesn't matter what the form of the resulting signal is. It's still just ONE amplitude varying over time. The fact that it doesn't look like a sine wave is irrelevant.

It is very relevant, to the contrary.
The fact that the resultant signal is not a sine wave may imply that there may be some useful information between two sampling points.

The Nyquist principle only works if the signal behaves like a sine between two sampling points.
20th January 2020
#608
Gear Maniac

Quote:
Originally Posted by Scragend
But we didn't have an accurate reproduction of the analogue signal until digital came along! That's what most people hate about digital - it tells the truth unlike the various historic approximations of waveforms ingeniously stored on various physical media - each stage of production towards the finished media added distortion, saturation and rolled a little more top-end off. This happened to be pleasing to the human ear - it certainly isn't accurate. Making digital even more accurate won't necessarily make it sound "better" to the punter on the street - scientifically better perhaps - but musically?
You are right. Making digital more accurate will not necessary make it sound better. It is much a matter of taste.
I don't try to proove that 96 khz sounds better but that it sounds different. It may be more suited for some types of music and less for others.
20th January 2020
#609
Lives for gear

Quote:
Originally Posted by Tom Barnaby
It is very relevant, to the contrary.
The fact that the resultant signal is not a sine wave may imply that there may be some useful information between two sampling points.

The Nyquist principle only works if the signal behaves like a sine between two sampling points.
But we can deconstruc ANY signal in a finite number of sines (within a given bandwidth). THERE IS NO OTHER SIGNAL !!!!!!!!!!'!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Now go outside and buy a book.

And of course there is useful information between two sampling points. If you would take just one minute and try to understand what you have been told! You would get it: this information is stored in the data ! It is reconstructed by the DA converter, only bandwidth limited !!!!!!

Only information above half the samplerate is lost, within the audio band NOTHING is lost.

N O T H I N G
20th January 2020
#610
Gear Maniac

Quote:
Originally Posted by Yannick
But we can deconstruc ANY signal in a finite number of sines (within a given bandwidth). THERE IS NO OTHER SIGNAL !!!!!!!!!!'!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Now go outside and buy a book.

And of course there is useful information between two sampling points. If you would take just one minute and try to understand what you have been told! You would get it: this information is stored in the data ! It is reconstructed by the DA converter, only bandwidth limited !!!!!!

Only information above half the samplerate is lost, within the audio band NOTHING is lost.

N O T H I N G
We can decontruct any signal in a finite number of sines. This is true. I do agree.

However, the sampling theorem is based on the sampling of sine waves. It works because the behaviour of the signal is predictable.
I showed above that the signal doesn't always behave like a sine wave between two sampling points.
Should we consider this as a problem ?
Could we proove that no useful information is lost in this case when sampling at standard rate ?
20th January 2020
#611
Gear Guru

Quote:
Originally Posted by Tom Barnaby
It is very relevant, to the contrary.
The fact that the resultant signal is not a sine wave may imply that there may be some useful information between two sampling points. The Nyquist principle only works if the signal behaves like a sine between two sampling points
The only information "between" two sampling points is by definition, ultra high-frequency information. Which has already been removed by the Nyquist filter. The signal always behaves "like a sine" because any signal - no matter how complex- is the sum of many sines.

Your attempts to force digital audio on to the Procrustean Bed of your limited imaginary picture of how it works will lead you only to the "stairsteps" fallacy or to the "ultrasonic hearing" woo-woo.

As others have already said, you really should educate yourself on this topic before you embarrass yourself further.
20th January 2020
#612
Lives for gear

Quote:
Originally Posted by Tom Barnaby
We can decontruct any signal in a finite number of sines. This is true. I do agree.

However, the sampling theorem is based on the sampling of sine waves. It works because the behaviour of the signal is predictable.
I showed above that the signal doesn't always behave like a sine wave between two sampling points.
Should we consider this as a problem ?
Could we proove that no useful information is lost in this case when sampling at standard rate ?
Where did you prove the signal does not always behave like a sine wave between two sampling points, sithin the bandwidth ?

Are you imagining things ?
20th January 2020
#613
Gear Maniac

Quote:
Originally Posted by Yannick
Where did you prove the signal does not always behave like a sine wave between two sampling points, sithin the bandwidth ?

Are you imagining things ?
You made me having a doubt, so I just made a listening comparison between a song recorded at 96 khz and another from a 44.1 khz CD. The more I listen to them and the more I hear the difference.

44.1 khz sampling means a sampling point every 20 ms or so . Is it right ?
20th January 2020
#614
Gear Guru

Quote:
Originally Posted by Tom Barnaby
It is very relevant, to the contrary.
The fact that the resultant signal is not a sine wave may imply that there may be some useful information between two sampling points.
No, it may not.

You already agree that any complex waveform is nothing more than added sine waves. Therefore the highest frequency that is within a complex waveform is a sine wave. Anything "between two sampling points" is filtered out. To the system it's non-existent.

The fastest rate of change is going to be determined by what the highest frequency sine wave looks like, not a complex wave. Because if something isn't a sine wave then it's a complex wave made up of sine waves.

Now, if you have two samples then any sine wave frequency below half of that sample rate can be accurately drawn. But so can any other complex waveform!. You can draw a sine wave that completes its cycle over 10 sample points, and you can draw one that completes over 1,000, or you can draw one that completes over just three. Between all options you can draw any number of complex waveforms as long as the highest frequency that makes it up is lower than half the sample rate.

Just grab a piece of paper with a grid on it and try it for yourself.
20th January 2020
#615
Gear Maniac

Quote:
Originally Posted by joeq
The only information "between" two sampling points is by definition, ultra high-frequency information. Which has already been removed by the Nyquist filter. The signal always behaves "like a sine" because any signal - no matter how complex- is the sum of many sines.

Your attempts to force digital audio on to the Procrustean Bed of your limited imaginary picture of how it works will lead you only to the "stairsteps" fallacy or to the "ultrasonic hearing" woo-woo.

As others have already said, you really should educate yourself on this topic before you embarrass yourself further.
Are you a graduated mathematician or electronics engineer ?
How can you be sure you exactly know how digital audio works and which problems are involved ?
You should try to trust your ears instead of relying so much on books.
20th January 2020
#616
Gear Guru

Quote:
Originally Posted by Tom Barnaby
You made me having a doubt, so I just made a listening comparison between a song recorded at 96 khz and another from a 44.1 khz CD. The more I listen to them and the more I hear the difference.

44.1 khz sampling means a sampling point every 20 ms or so . Is it right ?
20th January 2020
#617
Gear Maniac

Quote:
Originally Posted by mattiasnyc
No, it may not.

You already agree that any complex waveform is nothing more than added sine waves. Therefore the highest frequency that is within a complex waveform is a sine wave. Anything "between two sampling points" is filtered out. To the system it's non-existent.

The fastest rate of change is going to be determined by what the highest frequency sine wave looks like, not a complex wave. Because if something isn't a sine wave then it's a complex wave made up of sine waves.

Now, if you have two samples then any sine wave frequency below half of that sample rate can be accurately drawn. But so can any other complex waveform!. You can draw a sine wave that completes its cycle over 10 sample points, and you can draw one that completes over 1,000, or you can draw one that completes over just three. Between all options you can draw any number of complex waveforms as long as the highest frequency that makes it up is lower than half the sample rate.

Just grab a piece of paper with a grid on it and try it for yourself.
I don't think that the fastest rate of change of the signal is determinated by the highest audible frequency.
Let's take the example of a 20 khz sinewave. If another sinewave of same or comparable frequency comes one or two ms after, we will see a fluctuation of the signal whose frequency will correspond to a period of 1 or 2 milliseconds.

This doesn't correspond to a complete waveform cycle, but this fluctuation of the waveform may be audible.
20th January 2020
#618
Gear Guru

Quote:
Originally Posted by Tom Barnaby
I don't think that the fastest rate of change of the signal is determinated by the highest audible frequency.
I didn't write "audible". "Audible" relates to the auditory system. That's a different matter than whether or not we're accurately describing an analog waveform.

As for the actual rate of change: The fastest rate of change is something that is instantaneously different. So for example the shape of a square wave. How do you get a theoretically perfect square wave? You can get one by adding an infinite amount of harmonic sine waves:

I.e. still looking at sine waves for the fastest rate of change.

Quote:
Originally Posted by Tom Barnaby
Let's take the example of a 20 khz sinewave. If another sinewave of same or comarable frequency comes one or two ms after, we will see a fluctuation of the signal whose frequency will correspond to a period of 1 or 2 milliseconds.

This doesn't correspond to a complete waveform cycle, but this fluctuation of the waveform may be audible.
If it isn't a sine wave it's a complex waveform. It can be broken down to sine waves. Is the sine wave above or below the allowed frequency? If it is below it can be represented. If above it will be filtered out.

Not sure why this is so hard to understand.
20th January 2020
#619
Lives for gear

Quote:
Originally Posted by Tom Barnaby
You made me having a doubt, so I just made a listening comparison between a song recorded at 96 khz and another from a 44.1 khz CD. The more I listen to them and the more I hear the difference.

44.1 khz sampling means a sampling point every 20 ms or so . Is it right ?
One thousand time less. If you even cannot think in the right orders of magnitude, you need to go outside and buy some books !
20th January 2020
#620
Gear Guru

Quote:
Originally Posted by Tom Barnaby
Are you a graduated mathematician or electronics engineer ?
No, but quite obviously neither are you.

The real difference between us is that I have studied what those mathematicians and EEs say about digital audio and clearly you have not. I decided against making up my own pseudo-science based on my imaginary metaphors and instead learned from the people who actually know. See?

FWIW, I have a Masters in Educational Technology, which simply means I have learned how to learn technical subjects.

Quote:
How can you be sure you exactly know how digital audio works and which problems are involved ?
I am a professional recording engineer. At one time, I harbored the same kinds of whimsical misconceptions about digital audio that you now exhibit. But being a professional, I felt it was important to me to learn exactly what was going on with the tools I use every day and how things like Nyquist really works.

So I studied. I read books. You can also watch YouTubes if you like.

We are not a bunch of cavemen poking at a walkie-talkie. Digital Audio did not fall off the loading ramp of a flying saucer. It was invented by human beings and (some) human beings understand it completely. I took the time and trouble to learn what those humans had to say about it. Pretty simple really.

You may 'hear' a difference. But you have no basis to attribute that difference to your own fanciful imaginings about baloney like what lies "between" the samples. What you hear could be as simple as the performance of the hardware converters, the filters, or yes, plain old confirmation bias. It is one thing to criticize the performance of real-world hardware etc. and it is something totally different to make up your own "science" about sampling theory.

It's OK to switch on your computer and simply use digital audio every day, not truly understanding how it works. Just like it is OK for your grandmother to start her car every day, not truly understanding how an internal combustion engine works. But your grandmother doesn't go on automotive forums and presume to tell the professional mechanics there how their car's performance issues are due to the "hamsters getting tired" .

Quote:
You should try to trust your ears instead of relying so much on books.
Oh, we sure wouldn't want to involve those nasty books

I think we are done here.

Or at least you are.

Last edited by joeq; 21st January 2020 at 12:12 AM..
20th January 2020
#621
Gear Maniac

Quote:
Originally Posted by mattiasnyc
I didn't write "audible". "Audible" relates to the auditory system. That's a different matter than whether or not we're accurately describing an analog waveform.

As for the actual rate of change: The fastest rate of change is something that is instantaneously different. So for example the shape of a square wave. How do you get a theoretically perfect square wave? You can get one by adding an infinite amount of harmonic sine waves:

I.e. still looking at sine waves for the fastest rate of change.

If it isn't a sine wave it's a complex waveform. It can be broken down to sine waves. Is the sine wave above or below the allowed frequency? If it is below it can be represented. If above it will be filtered out.

Not sure why this is so hard to understand.

Do you agree with the idea that if the signal does not behave like a sine between two sampling points, then it wiil not be reconstructed properly from the digital file ?
20th January 2020
#622
Gear Guru

I just realized that the guy is a troll. He's been told ALL of this before and just ignored it. So we should probably just move along.
20th January 2020
#623
Gear Guru

Quote:
Originally Posted by Tom Barnaby
Do you agree with the idea that if the signal does not behave like a sine between two sampling points, then it wiil not be reconstructed properly from the digital file ?
START HERE
21st January 2020
#624
Gear Maniac

I think we should focus more on blind test procols and how the listeners should be prepared to hear potential audio differences between files.
If listeners know what kind of difference they may hear, shey should be able to discriminate between sample rates.
21st January 2020
#625
Gear Maniac

Quote:
Originally Posted by mattiasnyc
It looks like you are avoiding my question.

It is an important point and i would like to know your opinion.
21st January 2020
#626
Gear Guru

Quote:
Originally Posted by Tom Barnaby
It looks like you are avoiding my question.

It is an important point and i would like to know your opinion.
The answer has been given over and over and over.

You either don't listen or you don't learn. Regardless you behave exactly like a troll. There is no value in your posts because of that. So I'll just ignore you from now on. Hopefully other people will do the same.
21st January 2020
#627
Gear Maniac

Quote:
Originally Posted by mattiasnyc
I just realized that the guy is a troll. He's been told ALL of this before and just ignored it. So we should probably just move along.
Calling other people a troll is not nice. You are free to participate in the debate or not.
21st January 2020
#628
Gear Maniac

Quote:
Originally Posted by mattiasnyc
The answer has been given over and over and over.

You either don't listen or you don't learn. Regardless you behave exactly like a troll. There is no value in your posts because of that. So I'll just ignore you from now on. Hopefully other people will do the same.
This is up to you,
However I guess that you are seriously involved in digital audio and you can't ignore that thousands of engineers use 96 khz for recording because they think it sounds more authentic.

Would you call those people trolls ?
21st January 2020
#629

Quote:
Originally Posted by Tom Barnaby
You made me having a doubt, so I just made a listening comparison between a song recorded at 96 khz and another from a 44.1 khz CD. The more I listen to them and the more I hear the difference.

44.1 khz sampling means a sampling point every 20 ms or so . Is it right ?
I believe you. The more YOU listen to them, the more YOU hear a difference.

Man can make him himself believe/hear anything.
21st January 2020
#630
Gear Maniac

Quote:
Originally Posted by coolbass
I believe you. The more YOU listen to them, the more YOU hear a difference.

Man can make him himself believe/hear anything.

Do you think that thousands of audio engineers hear things that do not exist ?
Topic:

## Welcome to the Gearslutz Pro Audio Community!

###### Registration benefits include:
• The ability to reply to and create new discussions
• Interact with VIP industry experts in our guest Q&As
• Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD \$20/year
• Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.