info@tomflair.com

Interesting! That means I'm 100 % right then and proof that I'm right. The sound can't be 100 % correct capture in example 96kHz sample frequency then 44.1 kHz, due to speed of the sample frequency "speed accuracy tradeoff"-curve. That means that the 44.1 kHz is 100% correct,= “44.1 kHz”-sound, but we still hear & capture more “magic” in 96 kHz or 192 kHz even if the “96kHz”-sound isn't 100 % correct. This is what we all call overtone, transients-tones, that do exist = that is why we hear and feel that it sounds better in higher frequency's!

Freddie H


so freddie - you are saying dan lavry is wrong and you are right cos you heard it ??? :-)
 
btw.. where are you located - i d really lopve to do the tests with you ....
 
and for the converters: i am not sure if those digi things can hold up with the real high end stuff like prism or weiss (and that other firm i just cant remmeber the name)   also single  specs on a sheet of paper are not that relevent - 120 dynamic range is nice - but looks like humans cant even percieve this dynamic range :-) whats mor important in the converter is the clocking, the jitter - the analog stage befor the ad chip etc..etc..
 
but - freddi i dont wanna argue with you, so if you "like" it better - its ok !
 
but there are no overtones above 26 k that interest humans in any way ... just ttake a look at a spectrograph and you will have the simple proof
 
 
cheers
 
 

Clock jitter is another factor in the studio too... dogs and cats can hear above 26k. Still in "real life" "sound", way over.
Like leaf in the forest or the wind blowing, water... and so on... evne though we can't capture or measure it, but it is still there..
 
 
Regards
Freddie
 
 

recorded

info@tomflair.com


pollux - what are you talking about? so are you trying to negate that samplerate is there to represent frequencies , and the higher the rate the higher the possible recorded frequency ????
obviously you cant represent very high freqs if your sampled "snippet" is longer than what you wanna represent (in digital its even doubled) BUT - really what is your point ???
this whole discussion is getting pretty ridiculous cos some people state act like they "knew" what they are talking about ....we have so many postings here talking about "dynamic range" for example -  as bitflipper pointed out it has NOTHING to do with sample rate - btw - your term "accuracy" stands for nothing else than for "wider frequency range" so please dont get me wrong - but dont correct me with posting wrong stuff or mixing up things ....
 
this whole discussion is futile:
 
i state this again: on a double randomized blind test NO ONE will be able to tell the differences between  any real single recorded acoustic signal  at 48 or higher if the converter is of good quality - this tests have all be done - all bat-ears FAILED ...EVER!
 
and i have posted it 4 times now that vsti /vst and external fx processing is a different thing - and i also explained why...
 
so praise the snake oil - cool your cables - put some gems on your speakers (dont forget the piece of toilet paper wrapped around the high mambrane or your ns10 and only work at 192khz when its full moon - your tracks will sound amazing - and dont forget to change your whole equippment every 2 yeras cos - it gets internally consumed by voltage and sounds weaker and weaker every day - guess i will open a esoteric audio-gear firm - surley it will make me richer
 
 

I guess you fall in the category you are describing then. Please check again how audio is digitized and restored back from digital.
It's not sampling "snippets", but making instant measures at the sampling rate frequency. for example 48 KHz means 48.000 measures a second.
This instant values are then used for building a wave form, just like the "draw the line following the numbers" games for kids.
The more samples you have in the same period of time, the more accurate the rendered waveform will be. This accuracy is non-linear, diminuishing as the sampled frequency approaches the samplig rate, simply because as the sampled frequency rises, there are  less instant measures that can be taken.
Also, "normal" audio is much more complex than a plain sine, so there are more things to take in consideration than the plain frequency of the incoming signal.

As I said, you can digitize frequencies above the sampling rate, but they will not be recorded accurately. If we are talking about a sine in the input, the output will be something else. 

From an "audio" perspective, this is of little relevance because the human ear cannot hear frequencies that high, yet you will still introduce more artifacts when digitizing or restoring at 48 KHz than at 96 KHz. As you said, this difference is not audible, but it does exist.

Anyway, I don't see the point on bashing Freddie for what he said. He said "I like to work this way and I will never go back".. fine.. He's not saying "you are all a bunch of caveman morons working at such low samplig rates.. I shall send my troops to conquer the world and force every single DAW user to work in 96KHz". I do dissagree with him on other topics, but he's not being treated fair on this one.

So why start yet again another "my <<put whatever you want here>> is bigger than yours" war?
96 KHz works for him, good.. 48 KHz works for me.. fine too.. I'm not trying to convince him that he's wasting his precious disk space and processor ressources... 

grab a cup of coffee (or a beer depending on your timezone) and calm down a bit whilst talking about something constructive..  ;)

info@tomflair.com


so just take out the tension of the technical side....probably the next thing that is "best" is some self programming sequencer taht is fed with all the best songs of all ages and that will always compose perfect songs itself - 

there are some generative music programs that are kinda interesting.. they are probably more talented musicians than me  

Nyquist frequency

MatsonMusicBox

pollux

Please check again how audio is digitized and restored back from digital.

It's not sampling "snippets", but making instant measures at the sampling rate frequency. for example 48 KHz means 48.000 measures a second.
This instant values are then used for building a wave form, just like the "draw the line following the numbers" games for kids.
The more samples you have in the same period of time, the more accurate the rendered waveform will be. This accuracy is non-linear, diminuishing as the sampled frequency approaches the samplig rate, simply because as the sampled frequency rises, there are  less instant measures that can be taken. 

sorry - this is just plain wrong - you don't understand how digital conversion works. Any two sampling rates will produce the EXACT same waveform as long as they are both over the Nyquist frequency for what is being sampled. The "old textbook" view that digital audio is converted like playing "connect the dots" and the more dots you have the "smoother" the result comes out is incorrect. That's simply not how it works.

No it's not wrong, there's an algorythm that converts the samples to waveforms, just like "connect the dots", excepting that it's using mathematical interpolation instead of a pencil.
These algorythms are able to reconstruct the waveform from the stream of samples without aliasing as long as the frequency of the waveform is under the Nyquist frequency (the half of the sample rate frequency BTW)

The Nyquist - Shannon sampling theorem says:
"...The continuous signal varies over time (or space in a digitized image, or another independent variable in some other application) and the sampling process is performed by measuring the continuous signal's value every T units of time (or space), which is called the sampling interval. In practice, for signals that are a function of time, the sampling interval is typically quite small, on the order of milliseconds, microseconds, or less. This results in a sequence of numbers, called samples, to represent the original signal. Each sample value is associated with the instant in time when it was measured. The reciprocal of the sampling interval (1/T) is the sampling frequency denoted fs, which is measured in samples per unit of time. If T is expressed in seconds, then fs is expressed in Hz.
Reconstruction of the original signal is an interpolation process that mathematically defines a continuous-time signal x(t) from the discrete samples x[n] and at times in between the sample instants nT..."

pollux

Nyquist frequency

MatsonMusicBox

pollux

Please check again how audio is digitized and restored back from digital.

It's not sampling "snippets", but making instant measures at the sampling rate frequency. for example 48 KHz means 48.000 measures a second.
This instant values are then used for building a wave form, just like the "draw the line following the numbers" games for kids.
The more samples you have in the same period of time, the more accurate the rendered waveform will be. This accuracy is non-linear, diminuishing as the sampled frequency approaches the samplig rate, simply because as the sampled frequency rises, there are  less instant measures that can be taken. 

sorry - this is just plain wrong - you don't understand how digital conversion works. Any two sampling rates will produce the EXACT same waveform as long as they are both over the Nyquist frequency for what is being sampled. The "old textbook" view that digital audio is converted like playing "connect the dots" and the more dots you have the "smoother" the result comes out is incorrect. That's simply not how it works.

No it's not wrong, there's an algorythm that converts the samples to waveforms, just like "connect the dots", excepting that it's using mathematical interpolation instead of a pencil.
These algorythms are able to reconstruct the waveform from the stream of samples without aliasing as long as the frequency of the waveform is under the Nyquist frequency (the half of the sample rate frequency BTW)

The Nyquist - Shannon sampling theorem says:
"...The continuous signal varies over time (or space in a digitized image, or another independent variable in some other application) and the sampling process is performed by measuring the continuous signal's value every T units of time (or space), which is called the sampling interval. In practice, for signals that are a function of time, the sampling interval is typically quite small, on the order of milliseconds, microseconds, or less. This results in a sequence of numbers, called samples, to represent the original signal. Each sample value is associated with the instant in time when it was measured. The reciprocal of the sampling interval (1/T) is the sampling frequency denoted fs, which is measured in samples per unit of time. If T is expressed in seconds, then fs is expressed in Hz.
Reconstruction of the original signal is an interpolation process that mathematically defines a continuous-time signal x(t) from the discrete samples x[n] and at times in between the sample instants nT..."

No time to argue endlessly on this one - suggest you read Digital Audio Explained by Aldrich. In short - (assuming we're above Nyquist frequency) sample points can yield one and only one resulting curve when converted. It's a mathematical truth. The "connect-the-dots" model that, unfortunate, is often used by amateurs in "white papers" and  other grade-school explanations that you find all over ARE SIMPLY NOT CORRECT. That is NOT how it works.

 

I don't know, Ken. If people start actually reading books 'n stuff, what will we then have left to argue about? Universal health care? Boring. Half the folks on this forum live in places that already have it. No, ignorance is unparalleled as a conversation-starter and should be nurtured for the sake of entertaining discourse.

pollux

recorded

info@tomflair.com


pollux - what are you talking about? so are you trying to negate that samplerate is there to represent frequencies , and the higher the rate the higher the possible recorded frequency ????
obviously you cant represent very high freqs if your sampled "snippet" is longer than what you wanna represent (in digital its even doubled) BUT - really what is your point ???
this whole discussion is getting pretty ridiculous cos some people state act like they "knew" what they are talking about ....we have so many postings here talking about "dynamic range" for example -  as bitflipper pointed out it has NOTHING to do with sample rate - btw - your term "accuracy" stands for nothing else than for "wider frequency range" so please dont get me wrong - but dont correct me with posting wrong stuff or mixing up things ....

this whole discussion is futile:

i state this again: on a double randomized blind test NO ONE will be able to tell the differences between  any real single recorded acoustic signal  at 48 or higher if the converter is of good quality - this tests have all be done - all bat-ears FAILED ...EVER!

and i have posted it 4 times now that vsti /vst and external fx processing is a different thing - and i also explained why...

so praise the snake oil - cool your cables - put some gems on your speakers (dont forget the piece of toilet paper wrapped around the high mambrane or your ns10 and only work at 192khz when its full moon - your tracks will sound amazing - and dont forget to change your whole equippment every 2 yeras cos - it gets internally consumed by voltage and sounds weaker and weaker every day - guess i will open a esoteric audio-gear firm - surley it will make me richer

I guess you fall in the category you are describing then. Please check again how audio is digitized and restored back from digital.
It's not sampling "snippets", but making instant measures at the sampling rate frequency. for example 48 KHz means 48.000 measures a second.
This instant values are then used for building a wave form, just like the "draw the line following the numbers" games for kids.
The more samples you have in the same period of time, the more accurate the rendered waveform will be. This accuracy is non-linear, diminuishing as the sampled frequency approaches the samplig rate, simply because as the sampled frequency rises, there are  less instant measures that can be taken.
Also, "normal" audio is much more complex than a plain sine, so there are more things to take in consideration than the plain frequency of the incoming signal.

As I said, you can digitize frequencies above the sampling rate, but they will not be recorded accurately. If we are talking about a sine in the input, the output will be something else. 

From an "audio" perspective, this is of little relevance because the human ear cannot hear frequencies that high, yet you will still introduce more artifacts when digitizing or restoring at 48 KHz than at 96 KHz. As you said, this difference is not audible, but it does exist.

Anyway, I don't see the point on bashing Freddie for what he said. He said "I like to work this way and I will never go back".. fine.. He's not saying "you are all a bunch of caveman morons working at such low samplig rates.. I shall send my troops to conquer the world and force every single DAW user to work in 96KHz". I do dissagree with him on other topics, but he's not being treated fair on this one.

So why start yet again another "my <<put whatever you want here>> is bigger than yours" war?
96 KHz works for him, good.. 48 KHz works for me.. fine too.. I'm not trying to convince him that he's wasting his precious disk space and processor ressources... 

grab a cup of coffee (or a beer depending on your timezone) and calm down a bit whilst talking about something constructive..  ;)

 
 
Exactly, thanks my friend!
I have no problem with that people disagree with me.
It is as it should be otherwise what's the point discussing it, if everyone think exactly the same, right!
 
 
Best Regards
Freddie

info@tomflair.com


god... with snippets i meant the size of time between two sapmled values - since i  am not a native english guy sometimes its hard to write espacially about technical stuff, b- also i never said that technically there is no difference between sampling rates - just that the difference is not relevant to human physiology...
sorry - maybe we jsut misinterpreted each other -
and i am especially not an esoteric or "mine is (obviously ;-) ) longer than yours guy)

its just this extreme tech-spec frenzy that is taking over in the last years that makes me a bit astounded...

also we are not bashing freddie - but since he really presents himself as the messiaha of "faster, newer, better" some other guys like me just feel to point out that all this specs-war might only optimize your work in a very small percentage (maybe) - and i think that its not optimal do degrade musicproduction into a specswar (even if i like gear- as i said)
its the passion, the experience, the material, and all the other stuff that makes a good song - and YES - me and a lot of other folks have produced "fat, shiny, crispy, detailed, full and crunchy" tracks also years ago - all at 44/16 with consoles that had 20 bit converters and 32 bit mix engines ....
so just take out the tension of the technical side....probably the next thing that is "best" is some self programming sequencer taht is fed with all the best songs of all ages and that will always compose perfect songs itself - but here it should at least a little ba about art ( i know- a big word) about individual sound and mood - not about (mostly) bits, bytes, specs and cash)

and finally again and again and again  - you probably know the tests - they showed the truth of our hyper ears....they do not even check what is better statistically over those who just guessed - which is funny and sad at the same time

regards

Agree on its not gear you have, it how you use it! A great song is a great song!
I have never presents myself as the messiaha of "faster, newer, better". I think you get me wrong! 
64bit is great, that doesn't make me a tech geek or a "gearslutz". I still believe in technology that actullay benefits for something more then just looking for a great specifications on gears.. I'm not into that specifications bag at all!
 
 
Regards
Freddie
 

Actually , it's not really "connect the dots". What happens in a DAC is you have a "stair stepped" voltage output (or perhaps a switching back and forth from zero to the sample voltages), which when passed through the analog filter, results in a smooth curve, that, in theory, matches the "original" signal perfectly up to the Nyquist frequency. In reality, imperfect filters mean the signals don't match exactly. But with modern oversampling converters using good filter design, filter artifacts should be inaudible.

In other words, the "connecting the dots" is performed entirely by the filters. And adding more "dots" (i.e. a higher sample rate), does not increase accuracy, because those dots can only contain "new" data above the Nyquist frequency.

When talking about "connecting the dots" or "impulse response", people try (often unwittingly) to sneak frequencies above Nyquist into the discussion, then use this to make false claims about how higher SR's are more "accurate" even below Nyquist. But except for the filter artifacts around Nyquist, higher SR's really aren't any more accurate.

drewfx

bitflipper


I don't know, Ken. If people start actually reading books 'n stuff, what will we then have left to argue about? Universal health care? Boring. Half the folks on this forum live in places that already have it. No, ignorance is unparalleled as a conversation-starter and should be nurtured for the sake of entertaining discourse.

Plus eleventy jillion times infinity or somethin.

In 15-16 century you were ignorant if you believe that the Earth and world was round.
Today, we know better!

Don't believe everything just because you have read it or some scientist say so. It true until someone prove that he or she was wrong. Einstein never listen to the crowed, and he was a great Scientist, even though he has been proven wrong on some points too.

Think about it.

Regards
Freddie

It true until someone prove that he or she was wrong.

Can't argue with that logic, Freddie!

Freddie

can you post a link to some of your work...I would like to hear it

thanks

Wiz

Wiz


Freddie

can you post a link to some of your work...I would like to hear it

thanks

Wiz

I don't like to talk about what projects and Artist I have "done" and what Artist I work with "right now". I'm very secret about my work, who and what I work with right now. I have done some movie & trailer projects too...
But I can post some old goodies that have done and some new one that I co write on...
 
 
Here some of the Artist/ Friends that I have worked with and some songs I produced and wrote...   
  

 
 
http://www.youtube.com/watch?v=6IxCaq3sX6s

http://www.youtube.com/watch?v=LFKGumJxz4U&
 
http://www.youtube.com/watch?v=pYImEf_-km0&feature=PlayList&p=AF22DCEC66A4DE23&index=0
 
http://www.youtube.com/watch?v=n9WV894Lh-w&feature=related   
 
http://www.youtube.com/watch?v=DrRC7Xds6l4&feature=PlayList&p=3A70B0D9B6BE3739&index=1

http://www.youtube.com/watch?v=IaDEDYnoiio&feature=related
 
http://www.youtube.com/watch?v=Ah5vVl2N17M&feature=related 
  
 
[link=http://www.youtube.com/watch?v=IN2A1qiR0tU] 
  
  
 
and more...