24bit vs 16bit, the myth exploded!

Sep 18, 2013 at 4:37 PM Post #1,306 of 7,175
Here is an interesting and highly entertaining video on the subject. 
http://www.youtube.com/watch?v=cIQ9IXSUzuM#t=276
 
I also recommend Ethan Winers 'The Audio Expert', all of his videos on youtube as well as his articles (for example on audio perception) to clarify common beliefs. The thing is he provides Test tracks and stuff, so you can test yourself rather than loose yourself in endless descussions. 
 
Sep 18, 2013 at 5:34 PM Post #1,307 of 7,175
  Here is an interesting and highly entertaining video on the subject. 
http://www.youtube.com/watch?v=cIQ9IXSUzuM#t=276
 
I also recommend Ethan Winers 'The Audio Expert', all of his videos on youtube as well as his articles (for example on audio perception) to clarify common beliefs. The thing is he provides Test tracks and stuff, so you can test yourself rather than loose yourself in endless descussions. 

 
This video has been mentioned many times here, a link to it should be a sticky, it is a very useful and extremely understandable explanation.
 
Sep 18, 2013 at 7:39 PM Post #1,308 of 7,175
   
This video has been mentioned many times here, a link to it should be a sticky, it is a very useful and extremely understandable explanation.

 
There is a minor detail in the explanation about the stairstepped digital waveform appearing as a perfect copy of the analog signal. The theory says that only a perfect reconstruction filter at the output of zero-hold process is required to produce the exact representation of the original sampled signal. Unfortunately, that filter has a impulse response of a perfect brickwall filter) and something that is not physically possible to design in real circuits. 
 
Sep 18, 2013 at 8:05 PM Post #1,309 of 7,175
   
There is a minor detail in the explanation about the stairstepped digital waveform appearing as a perfect copy of the analog signal. The theory says that only a perfect reconstruction filter at the output of zero-hold process is required to produce the exact representation of the original sampled signal. Unfortunately, that filter has a impulse response of a perfect brickwall filter) and something that is not physically possible to design in real circuits. 

 
Fair point, let's just call it a very good facsimile.
 
Sep 18, 2013 at 10:59 PM Post #1,310 of 7,175
That is a belief. The sooner you understand this, the sooner you can realise your method is neither better nor worse than others.


All opinions are not created equal. Some are supported by evidence and some are not.


My ears and what I hear are my evidence.

I can accept your belief system (Science) without insult.

Why do you empiricists lack the ability to reciprocate?
 
Sep 18, 2013 at 11:47 PM Post #1,311 of 7,175
  Quantization noise, of course, lots of it.
 
These were ambient environment nature recordings, dawn and dusk, far out in the countryside (KY).
 
As another member has suggested to me directly, a very quiet preamp with more available gain would have addressed this challenge before the A/D stage.

 
What you should have heard was hiss. Lots of it. With a little ambient nature noise mixed in.
 
The other member mentioned the most likely cause of the problem - insufficient gain. The loudest internal noise in the system should be the microphone self noise or preamp noise. If you close the microphone in a padded case to eliminate external noise, you need enough preamp gain to bring the recorded noise level up to -80 dBFS or so. 80 dB of dynamic range should be enough unless you're recording thunderstorms or space shuttle launches... (See below.)
 
I recall reading a piece by a "nature sounds" enthusiast some time ago, where he mentioned that it's getting very hard to find locations in the US where there are long periods (more than 10 minutes or so) of freedom from man-made noise such as traffic or aircraft.
 
I have the "Thunderstorm in the Rockies" disc by Mobile Fidelity, and it stretches the capabilities of vinyl. I also have Bob Katz's recording of a shuttle launch, in 24/96 4-channel. That's a little too dynamic for 16 bit.
 
Sep 19, 2013 at 1:30 AM Post #1,312 of 7,175
My ears and what I hear are my evidence.

I can accept your belief system (Science) without insult.

Why do you empiricists lack the ability to reciprocate?

This is off into Philosophy of Science.
 
There is no such thing as private evidence in science. If your interest is "my evidence"--and that's perfectly legitimate as an activity, but it has no bearing on Sound Science--then you have nothing with which to persuade others. The way to proceed persuasively here is to attempt to prove yourself wrong ("Can I actually hear the difference?") and fail to do so (SEE: Karl Popper).
 
Sep 19, 2013 at 2:10 AM Post #1,313 of 7,175
My ears and what I hear are my evidence. I can accept your belief system (Science) without insult. Why do you empiricists lack the ability to reciprocate?


Oh man! That is the EASIEST to answer! Because I can't hear with your ears... and I can't think with your brain. But if you want to make some points that apply to both of us, I'm happy to listen.
 
Sep 19, 2013 at 2:28 AM Post #1,314 of 7,175
Oh man! That is the EASIEST to answer! Because I can't hear with your ears... and I can't think with your brain. But if you want to make some points that apply to both of us, I'm happy to listen.


he walked right into that one
 
Sep 19, 2013 at 3:21 AM Post #1,315 of 7,175
Oh man! That is the EASIEST to answer! Because I can't hear with your ears... and I can't think with your brain. But if you want to make some points that apply to both of us, I'm happy to listen.

 
BOOM !
HEADSHOT
very_evil_smiley.gif

 
Sep 19, 2013 at 3:27 AM Post #1,316 of 7,175
  Here is an interesting and highly entertaining video on the subject. 
http://www.youtube.com/watch?v=cIQ9IXSUzuM#t=276
 
I also recommend Ethan Winers 'The Audio Expert', all of his videos on youtube as well as his articles (for example on audio perception) to clarify common beliefs. The thing is he provides Test tracks and stuff, so you can test yourself rather than loose yourself in endless descussions. 

 
this is such an awesome video.  
 
Sep 19, 2013 at 6:49 AM Post #1,317 of 7,175
  Fair point, let's just call it a very good facsimile.

 
It can be very good if the frequency is not too close to Fs/2. For example, at 44100 Hz sample rate, an impulse response length of only 2 ms is enough to attenuate the 24100 Hz image of a 20 kHz tone by more than 120 dB. Although the oversampled signal still needs to be reconstructed with an analog filter, so there will still be some images, but far outside the audio band, and at a relatively low level.
 
Sep 19, 2013 at 11:55 AM Post #1,318 of 7,175
   
There is a minor detail in the explanation about the stairstepped digital waveform appearing as a perfect copy of the analog signal. The theory says that only a perfect reconstruction filter at the output of zero-hold process is required to produce the exact representation of the original sampled signal. Unfortunately, that filter has a impulse response of a perfect brick wall filter) and something that is not physically possible to design in real circuits.

 
in theory the perfect reconstruction filter likes "perfect" weighted Dirac Impulse train "analog" input
 
 
a zero order hold has in band frequency response roll off and requires EQ even with a perfect reconstruction filter
 
 
of course zero order hold is a lot easier to produce with our current electronics - so you always see the stair step waveform in popularized explanations
 
Sep 19, 2013 at 1:23 PM Post #1,319 of 7,175
   
in theory the perfect reconstruction filter likes "perfect" weighted Dirac Impulse train "analog" input
 
 
a zero order hold has in band frequency response roll off and requires EQ even with a perfect reconstruction filter
 
 
of course zero order hold is a lot easier to produce with our current electronics - so you always see the stair step waveform in popularized explanations

 
But I guess nobody produces a traditional binary-weighted DAC anymore, so this issue is somewhat moot. Delta-sigma is the way.
 
Sep 19, 2013 at 5:43 PM Post #1,320 of 7,175
even the last generation of full resolution "flagship" grade audio ladder DACs were advertised for at least 8x oversampling
 
so upsampling with digital filtering, correction were here already
 
but yes multi-bit delta sigma seem to have won the market
 
but there are a few hold outs claiming delta-sigma noise shaping loop dynamics, "noise floor modulation" is evidence of some mysterious musical soul destroying "time domain error" that engineers are overlooking
 

Users who are viewing this thread

Back
Top