Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: 16 bit audio doesn't have 105db dynamic range? (Read 12637 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

16 bit audio doesn't have 105db dynamic range?

hi learned people,

need some help here, i'm a novice with not a huge grasp on the math regarding sampling theory, DSP's, etc... (but am a keen enthusiast willing to learn    )

someone has said this to me and i have no clue if it's right or wrong; any views please? thanks, phil

"16 bit audio doesn't have 105db dynamic range. Increasing the dynamic range by dithering is a mathematical trickery that only works by averaging long periodic signals like pure sine waves over many cycles; real music is essentially non-periodic, think of percussion"

16 bit audio doesn't have 105db dynamic range?

Reply #2 –
The RMS noise amplitude of a properly dithered 16-bit signal is -96.3 dB, so you get a max signal-to-noise ratio of 96.3 dB. This improves to 98.1 dB for sine waves.

The perceived dynamic range, especially when using noise shaping, is much higher. Up to roughly 120 dB. (Simply push your favorite signal down to -110 dB, quantize with noise shaping to 16 bits, then amplify and listen ... you will still hear your signal.)


Quote
"16 bit audio doesn't have 105db dynamic range. Increasing the dynamic range by dithering is a mathematical trickery that only works by averaging long periodic signals like pure sine waves over many cycles; real music is essentially non-periodic, think of percussion"

That is wrong.
First of all, this is no mathematical trickery. Dither simply decorrelates quantization errors from the input signal. It is just added noise, and as such doesn't care at all about the periodicity of the input signal.
Something like the initial hit of a cymbal is still several tens of samples too long, before it would simply disappear in the dither.

In fact, quantizing a low level sine without dither will cause lots of distortion, which sounds nasty. With "real music" this is even worse, since low-level details will simply be rounded to zero (silence) or at least very distorted.
"I hear it when I see it."

16 bit audio doesn't have 105db dynamic range?

Reply #3 –
LP12, the person who said that to you about dither seems to be saying that despite being "trickery," dither does work, really, but just not for percussion?

Without worrying about the theory behind it, this is easy enough to test. In an audio editor, reduce the amplitude of a 24-bit file by 96 dB. (It's fine if it started as 16-bit and was saved as 24 before reducing the volume). Now you know the audio is in the -96 to -108 dB range. Dither when converting back to 16-bit. Play the file and see what you hear. You probably need to crank up the volume by 60 dB or so to get it far enough above the ambient noise of your listening environment to be audible.


...I hear percussion, including the claps which I assume begin with the type of short-duration sounds he's concerned about. Is it crisp and clear? Not really, I mean it's "lossy"... and maybe that's what was meant, but it is wrong to say the dynamic range isn't there.

(Suggestions for improving on this test are welcome. I used Audition, with triangular shaped dither with default settings).

16 bit audio doesn't have 105db dynamic range?

Reply #4 –
thanks for the helpful responses people; xnor i've used your reply in one of the linn forums, hope that's ok...

16 bit audio doesn't have 105db dynamic range?

Reply #5 –
Oh Linn... you gotta keep in mind that their agenda is to convince as many people as possible that their "hi-res" tracks, which they sell for a premium, sound better. They don't care about using misinformation to achieve this.
They demonstrated this clearly in an incredibly dishonest "format comparison" test, where they used a different (better) master for the 24/96 file.

I wouldn't be surprised if they deliberately botched their 16/44.1 files. Never bought from them, never will.
"I hear it when I see it."

16 bit audio doesn't have 105db dynamic range?

Reply #6 –
Quote from: xnor on
Never bought from them, never will.
Their CDs were very good.
From snippets on their website, I don't think all their current releases are really up to "audiophile" standards; on some they use dynamic range compression(!).


When carrying on a discussion across two forums, it's usual to give a link.
http://forums.linn.co.uk/bb/showthread.php...29701#pid329701

It's a discussion starting from a comment in Monty's article.
https://xiph.org/~xiphmont/demo/neil-young.html


It didn't really go anywhere.

Cheers,
David.

16 bit audio doesn't have 105db dynamic range?

Reply #7 –
Quote from: 2Bdecided on
Their CDs were very good.
From snippets on their website, I don't think all their current releases are really up to "audiophile" standards; on some they use dynamic range compression(!).

The important question is whether they use it consistently or if they botch 16/44.1 so they can earn more money with 24/96.
"I hear it when I see it."

16 bit audio doesn't have 105db dynamic range?

Reply #8 –
There are quite a few demos out there on the web that show how dither, especially noise shaped, improves the perceived dynamic range of very low bit depth music, not just sine waves. I think I understand the point of the question, though. Perhaps we can say that dither (esp. noise shaped) *genuinely* improves the dynamic range of real music up to a certain bandwidth, or that its improving effect reduces as signal bandwidth increases..? Effectively, if we dither at ultrasonic frequencies, it's just another form of oversampling is it not?

Quote from: xnor on
The important question is whether they use it consistently or if they botch 16/44.1 so they can earn more money with 24/96.

This is a specific example of a wider irony: every time the audiophile brotherhood casts aspersions on 'consumer level' digital audio in general (and you'll even see articles on it in the daily papers), they reinforce in the public's mind that the music they themselves consume is disposable pap, never to be paid for, never to be valued.

16 bit audio doesn't have 105db dynamic range?

Reply #9 –
Quote from: Green Marker on
Perhaps we can say that dither (esp. noise shaped) *genuinely* improves the dynamic range of real music up to a certain bandwidth, or that its improving effect reduces as signal bandwidth increases..?

Well yes, kinda. At low SPL, our hearing perceives noise in the kHz range much louder, so if you shape the noise out of this range the noise will sound less loud, less bothersome.
Also, with noise shaping you can afford using less added dither.


Quote
Effectively, if we dither at ultrasonic frequencies, it's just another form of oversampling is it not?

Oversampling is just sampling at a higher sampling rate than the Nyquist rate, but I guess you're referring to something like delta-sigma, which makes heavy use of noise shaping.

Well, you can shift a lot of energy to above 20 kHz, even with just 44.1 kHz sampling rate, but then you have lots of energy up there which you probably want to remove during playback ... else you might damage your tweeters or other components.
"I hear it when I see it."

16 bit audio doesn't have 105db dynamic range?

Reply #10 –
did any one else look at mjb2006's  example above,  with a wave editor, - just three levels -1,-0,+1 . Pretty impressive if you haven't seen this particular  visual demonstration of dither in digital audio before.

16 bit audio doesn't have 105db dynamic range?

Reply #11 –
Quote from: KMD on
Pretty impressive if you haven't seen this particular  visual demonstration of dither in digital audio before.


In fact if I use EQ to push down the top end of the spectrum, the whole thing sounds fairly normal.

I first experienced this when I set foobar's output to 8-bit (this was possible in Windows XP). I expected grave distortions, but what I got was a little bit of smooth background noise.

16 bit audio doesn't have 105db dynamic range?

Reply #12 –
Quote from: dhromed on
Quote from: KMD on
Pretty impressive if you haven't seen this particular  visual demonstration of dither in digital audio before.


In fact if I use EQ to push down the top end of the spectrum, the whole thing sounds fairly normal.

I first experienced this when I set foobar's output to 8-bit (this was possible in Windows XP). I expected grave distortions, but what I got was a little bit of smooth background noise.



Me, too.

That there science stuff is pretty cool, no? ;-)

16 bit audio doesn't have 105db dynamic range?

Reply #13 –
Darn straight, yessir.

16 bit audio doesn't have 105db dynamic range?

Reply #14 –
Here is a demo of 5-bit drum recordings with and without dither.

5-bit drum sounds

16 bit audio doesn't have 105db dynamic range?

Reply #15 –
So LP12, did any of this change any minds in your discussion on the other forum?
 

16 bit audio doesn't have 105db dynamic range?

Reply #16 –
Quote from: WernerO on
Here is a demo of 5-bit drum recordings with and without dither.

5-bit drum sounds

Link doesn't work for me:
Code: [Select]
400 Bad Request

Your browser sent a request that this server could not understand.
Apache/2.2.15 (Red Hat) Server at www.audiochrome.net Port 80