[mongolianfly]

Just a minor correction, 144db + a 50db noise floor does not equal 180db. That's like saying a few people talking at 60db each is as loud as a lawn mower.
 
Quote:

...144dB + say 50dB for the room's noise floor. But 180dB is the figure often quoted for sound pressure levels powerful enough to kill and some people have been killed by 160dB.

 

[D. Lundberg]

 
Quote:

Increasing the dynamic range also decreases the difference between steps... 

 
Audio waveforms have four characteristics: frequency, amplitude, phase (relative to other waveforms) and dynamic range. The dynamic range is the only one that is affected by the bit depth.
Increasing the amount of quantization steps means less quantization error. Less quantization error means less quantization noise. Less quantization noise means lower noise floor. And lower noise floor means greater dynamic range.
 
"If Q is the number of quantization steps and D is the dynamic range then
 
Q = 2^(D/6). or D = 6 Log2 Q"[1]
 
And the noise floor can be shaped to suit your needs. That's why modern D/A and A/D-converters can work with just a few bits and still have >100dB of dynamic range 20Hz-20kHz.
 
Quote:

and as I said, I have, in experimentation, listened to an 8 bit audio signal and even "properly reconstructed" the voltage jumps were noticeable.

You won't hear any voltage jumps. If the signal is properly reconstructed (and properly dithered) you'll get the original signal and some amount of white background noise that may or may not be audible (depending on SPL, room, noise shaping, etc.) and/or mask parts of the signal.
And if it isn't properly reconstructed (truncation, bad or no reconstruction filter(s), etc.) you'll get a signal that is (to some degree) distorted, and that is probably what you heard.
 
[1] Aldrich, Nika. Digital Audio Explained: For The Audio Engineer, 2005, Sweetwater.

 

[D. Lundberg]

Quote:

Just a minor correction, 144db + a 50db noise floor does not equal 180db. That's like saying a few people talking at 60db each is as loud as a lawn mower.

I assume he meant that you'll need to play at a level exceeding 180dB SPL to hear every part of a signal with 144dB of dynamic range in a room with 50dB SPL background noise, and that is accurate.

 

[khaos974]

Are you sure about that? If I generate a -10 dB pink noise file and add a -15 dB 440 Hz sin wave to that file, won't I be able to hear the 440 Hz signal?
 
Quote:

I assume he meant that you'll need to play at a level exceeding 180dB SPL to hear every part of a signal with 144dB of dynamic range in a room with 50dB SPL background noise, and that is accurate.

 

[wyager]

@D. Lundberg
The entire first part of your post I agree with... You act like you are correcting me, but that's all that I am saying... Those equations you posted describe the relationship I am talking about, if D goes up then Q goes up. If Q goes up, the difference between steps becomes smaller.
 
The second part, about the signal being "properly reconstructed", I'm not so sure... Even if a "proper" signal is dithered, it still won't sound as good as the original. The amount of noise is going to be proportional to the quantization error you hope to replace, which is a lot at 8 bits.

 

[khaos974]

No one is saying than an 8 bit signal sounds as good as a 16 bit one, it's just that if it is properly reconstructed, there should be no jumps.
 
I remember a CD track dithered down to 2 bit once, the noise was as loud as the song itself, but there was some "jumping around", but it was far less drastic than I expected with a file with 4 possible values...
 
A visual example of dithering, in order, the original file (16.7 million colors possible), a 16 colors with withering, a 16 colors with no dithering.
 

 

[wyager]

Wait-in the second file, the cloud is still visible but in the non-dithered file it is not. If the second file was truly a dithered version of the third file, there would be no cloud, right? Are you sure you didn't dither first and compress second?

 

[khaos974]

To answer your question, all files were saved with the same jpeg compression, and I checked that they were visually very close to the uncompressed version in my photo editor.

That's the whole point of dithering actually, the whole range of white to blue values is not present in a 16 color file, the dithered file compensated this by alternating too white and too blue pixels to create a blueish white cloud, the undithered version simply truncates the color values to the nearest color.
 
And there are even multiple algorithms of dithering that 'decide" how to reduce bit depth the best possible way. The DSD coding of SACD has a 1 bit depth with some advanced dithering and noise shaping.

 

[wyager]

So then it was dithered before compression, not after, right? I believe audio dithering is different, the wave is dithered during decompression. Otherwise all you get is quantization error with extra noise.

 

[Canuckabroad81]

absolutely! 
I for one agree with your research. Sticking recordings on a 24bit setting as apposed to 16bit doesn't accomplish much unless you are mixing.
I will agree that once sound goes through a compressor or external equipment we start to degrade the audio quality, so the extra bit depth might help reduce the noise added to the audio waveform (transistor noise, RF transients...) such is the case with tube amps which warms up the audio via distortion.  The nice kind of distortion!  
I for one find digital almost all Solid state amps to be lacking in depth of sound, as if something appears to be missing. I guess that's what you Audiophiles call COLD sound? Am I right? 
 
 
 Sorry, I am a newbie. 

 

[khaos974]

Quote:

So then it was dithered before compression, not after, right? I believe audio dithering is different, the wave is dithered during decompression. Otherwise all you get is quantization error with extra noise.

 
I don't know what compression you are talking about, for the picture above, it was color reduced/dithered from the original uncompressed file, and then jpeg compressed.


 
Quote:

absolutely! 
I for one agree with your research. Sticking recordings on a 24bit setting as apposed to 16bit doesn't accomplish much unless you are mixing.
I will agree that once sound goes through a compressor or external equipment we start to degrade the audio quality, so the extra bit depth might help reduce the noise added to the audio waveform (transistor noise, RF transients...) such is the case with tube amps which warms up the audio via distortion.  The nice kind of distortion!  
I for one find digital almost all Solid state amps to be lacking in depth of sound, as if something appears to be missing. I guess that's what you Audiophiles call COLD sound? Am I right? 
 
 
 Sorry, I am a newbie. 

 
 
For modern audio, the signal is usually recorded at 24 bit undithered, since no A/D converter achieves real 24 bit depth due to thermal noise. The tracks are mixed/mastered inside a DAW is 32/64 bit precision. The output file is disthered down to 16 bit with some noise shaping.

 

[wyager]

Quote:

Quote:

So then it was dithered before compression, not after, right? I believe audio dithering is different, the wave is dithered during decompression. Otherwise all you get is quantization error with extra noise.

 
I don't know what compression you are talking about, for the picture above, it was color reduced/dithered from the original uncompressed file, and then jpeg compressed.

Think about it. If the third picture was dithered, there would be no cloud. If the first picture was dithered and THEN subjected to the same compression as the third picture, the cloud would be visible (and it is).
It's just that I don't think you really want that when reconstructing audio.

 

[khaos974]

The first pictured is "high resolution", the second (with the cloud) dithered down to 16 colors, the third (no cloud) is truncated down to 16 colors.
 The jpeg compression is invisible in all pictures.
 
It a photo that I took myself, the whole processing chain is known to me.

 

[wyager]

Yes, I got that... I'm just saying that what you call "dithering down" appears to be dithering then truncating versus truncating then dithering. I believe audio, however, is often truncated then dithered (dithering occurring at the time of playback).

 

[khaos974]

Hmmm, I actually don't know whether it's truncate then inject noise or inject noise then truncate, the former makes more sense to me, my understanding is that the term dither refers to both steps.
 
But the key point is also that dithering (truncate + inject noise) happens at the mastering stage, ie. the file on the CD/computer is already dithered, there is no dithering at the time of playback. The only situation where you dither at the time of playback id if you need to convert a 24bit file to 16 bit because your DAC doesn't support 24 bit.