[Aleatoris]

Quote:

Originally Posted by Rempert /img/forum/go_quote.gif If a mosquito sneezes in a noisy factory, has it contributed anything to the noise problem?

No, but I want to be able to hear that moquito sneeze, dammit.

 

[manaox2]

Quote:

Originally Posted by Aleatoris /img/forum/go_quote.gif No, but I want to be able to hear that moquito sneeze, dammit.

You hear it. You get malaria. Bill Gates will be happy to help. http://www.foxnews.com/story/0,2933,488348,00.html

 

[gregorio]

Quote:

Originally Posted by Rempert /img/forum/go_quote.gif If a mosquito sneezes in a noisy factory, has it contributed anything to the noise problem? Mastering engineers claim to hear differences between various dithering algorithms and choose the best one for that particular job. These noise-shaped dithering algorithms are referenced in the original post. How is the mastering engineer supposed to choose something he cannot hear? If the mastering engineer can hear coloration to the piece caused by the dither, surely an audiophile with top line gear and a treated listening room could as well. If coloration from the dither is audible, and distortion from truncation if one chooses not to dither even worse, then one must conclude that an audible difference can be heard between 24 bit and 16 bit audio without raising peak levels so high as to cause deafness or deathness. That dither is even mentioned in this discussion suggests an acknowledgment that 16 bits worth of signal to noise ratio is not foolproof. If 96 db were a great overkill, truncation distortion would be a purely academic topic and dither would be some irrelevant relic of 8 bit audio.

Good points but not quite how it works in practice. There are a few reasons for this but they are not so easy to explain without using some terminology:

1. Most applied dithering algorithms are incorporated into limiters. So when you master down to a distribution format you are not just applying dither and a re-quantising process but also applying compression with an infinte ratio (limiting). So differences between dithering programs also include the sonic qualities of the limiter, which in general are rather noticeable.

2. There are a number of dithering algorithms used and they all have slightly different properties and uses. For example, Type 1 Noise Shaped Dither is a dither algorithm which most strongly re-distributes the dither noise to extremes of the frequency spectrum where it is less percievable. This gives CD a perceptual dynamic range of 120dB. In other words, the dither noise would appear to the listener to be 120dB lower than the loudest noise in the music track. However, we only use this noise shaped dither as a final process, if we used this type of dither whenever we processed a channel in the mix and then sum those channels together, we are potentially going to hear the dither noise because it is an accumulation of dither noise concentrated a small frequency band. So in this case we would use a non-noise shaped dither algorithm where the noise is spead evenly across the whole frequency spectrum.

3. Sometimes, to check what is really happening, a small and very quiet section of the music is selected and this section is played back at high volume (with the amp whacked up!) to identify any artifacts near the noise floor. The point is that we can instruct the software to just playback the selection, so we avoid the risk of playing back much higher dynamic material which would possibly blow our monitors and/or our ears. So we often check what is near the noise floor even though we know the consumer won't be able to hear it. Not everyone does this but those of us with a lot of professional pride and who are really picky about SQ often do.

Compared to noise-shaped dither, truncation distortion could sound 30dB louder. The other point is that truncation distortion is not un-correlated noise like dither, this means that truncation distortion can act (modulate) with the program material!! So although truncating distortion going from 24bit to 16bit is very unlikely to be heard, it's affects on the program material may be noticeable.

Sorry this post was a bit more technical, hopefully though it answers your post?

G

 

[scompton]

Good thing I put in the caveat that I could be completely wrong

 

[Clutz]

Quote:

Originally Posted by scompton /img/forum/go_quote.gif Not that I disagree in general, but I thought law was applied only to those theories that were proved mathematically. Newton came up with calculus just so he could prove this theories. I could be completely wrong

I'm not certain this is true. If you prove a theory mathematically, all you've done is proven the mathematical validity of that theory. You could come up with a legitimate mathematical proof for a theory that is imperfect. There are many mathematical proofs for string theory, but all of those proofs are contingent upon the underlying physical understanding of the theory being correct. The math for the theory may be correct, but that doesn't mean it's a perfect representation of nature.

[Note: I didn't see mape00's response, otherwise I wouldn't have responded. I did not intend to belabor the point]

 

[CDBacklash]

Quote:

Originally Posted by gregorio /img/forum/go_quote.gif Sorry, but I'm afraid you have misunderstood how digital audio works. The bit depth in digital audio encodes the dynamic range, it is the sample rate that is responsible for the frequency range. 8bit 44.1kFs/s has exactly the same frequency range as 24bit 44.1kFs/s, both max out at 22,050Hz. This cannot be disputed as it is a basic tenet of the Nyquist Sampling Theorem. With lower bit depths (say 8bit) the dynamic range (48dB) might be such that the dither noise is noticeable enough to mask the high frequency content. But with 16bit the noise floor is so low that it cannot be heard and therefore cannot mask the high frequency content. So in fact when you say 8bit recordings have no high frequencies you are mistaken (unless there is also a lower sampling rate). 8bit 44.1kFs/s actually has more high frequency content than 16bit 44.1kFs/s because there is the program material plus 48dB more noise distributed throughout the 22,050Hz frequency range. I also think you need to read my original post to understand why the greater dynamic range of 24bit compared to 16bit is neither desirable nor even possible. Hope this clears things up a bit, G

Sampling rate does not eliminate highs it simply causes aliasing and misrepresentation of them. It is typical for encodes to apply a lowpass that eliminates the frequencies - and many will do this based on a combination of bit depth and sampling frequency. Typical sampling rates for 8 bit are 22.05khz, and not 44.1khz. PCM will accept 44.1 khz for 8bit audio, but it is hardly ever used: see game boys, the NES and the good old telephone. Sure, your numbers are correct in theory, but they are not default. As for your precious 16bit, a lot of studios are now recording in 24bit and then converting 16bit for CD media (oh no!).
You do know amplitude quantisation occurs in all digital media where each waveform is not represented as a complete picture, but merely as a snapshot, right? The resolution in 24bit audio is much higher and I am pretty sure you can get a much smoother representation of dynamic (note: dynamic, not contour) whereas noise will be added to lower bitdepths to snap each sample to the supported resolution. It's not like 24bit audio is just adding more volume onto the audio. There is the capacity to use a larger number of different amplitudes per sample! There are more ticks to snap to in a 24bit digital recording and it will be a more accurate representation of the sound if done correctly. You can have a "louder" signal from 24bit audio which can then be scaled down (a lovely thing called the volume knob), to a level that is not disimilar to the same song in a 16bit audio stream. However the bits will _not_ be the same. There will be greater dynamic accuracy in the 24bit stream because you can use the extra room up the top and then audibly reduce the distance between each quantisation with both encode and the volume knob...
As for the additional dynamic range of 24bit audio being undesirable, I think you are pressing it just a little to far with this statement. Welcome to the sound engineer. Many sound engineers are not 12 yearold kids who love dynamic range and are going to have passages that are terribly soft with drum hits that are amazingly loud. Most engineers will apply a brickwall (that is a compression filter with a small attack and release time) and then increase the volume of the track. My personal favourites will just record and slap it on tape. I doubt 24 bit audio will help us with regards to the loudness war. However, with regards to recording one shout and then putting it on media "as is" I think 24bit resolution is definitely worthwhile. Why would I want my drums to have less impact than they were intended to by the artists? Musicians do have ears.
Thread is becoming a bit monotonous and boring... If you are afraid of change, or cant afford or get high resolution gear/audio at the moment dont worry about it, the price will come down, and if things pan out the way they have for the last while with digital media, 24bit will come to the fore later on in this life. 16 bit audio is not perfect and as such it does not do justice to music, albeit it performs very well. Digital storage mediums will _never_ do audio true justice because of the quantisation that occurs. Even then one could argue that the warm hum you get from analog reproduction is not doing it justice either (although i particularly like it). 24bit audio has the capacity to store orchestra hits or heaven forbid cannon blasts (from a particular epic) far more realisticly because of the dynamic range without the overall sound becoming too loud.

We are also seeing a generation with large amounts of hearing loss due to pressing their volume switches upwards because thats what compressed music wants you to do. Have you ever noticed that you often turn down the volume on drums with good 'crack' to them? Perhaps, if the mainstream industry plays it fairly perhaps we may be able to avoid a generation of deaf people.

Whether or not you or I can personally hear a difference at this point in time, the fact of the matter is there is a difference between 16 and 24 bit audio and frankly you are starting to smell a bit like a troll. Then again, maybe everything I learned in college is false and the big bad internet has one-upped me again.
(sorry for bad english)
**the bit-depth war is getting to me just as much as the loudness war. Let better technology come all it wants. It's not going to affect you at all.

 

[Aleatoris]

uh guy, he IS a sound engineer.

 

[CDBacklash]

uh, so was I. If you were to have a large amount of loudness followed by a very quiet section, with enough clarity in your equipment (which is always getting better) you may be able to pick up on the noise floor quite easily, especially as the sound approaches -80dBFS.
What the OP wants you to believe is "great sound" is actually just a dither doing its job and there is nothing particularly special about that at all.
**what he is saying is nothing particularly new or exciting. This has been a hot topic over at HAF for a long time, partciularly the cannon in teldecs 1812 overture which live exceeds the dynamic range by enough that can be considered acceptible in 16bit audio (in my opinion), where noise may become audible.
**the main problem with with 24bit playback atm is that the SNR of most equipment in use at the moment is only slightly better than pure 16bit. You can stretch this through simple thermodynamics (that is, cooling the equipment such as treating it cryogenically) and get a much better noise resolution but as far as approaching 24bit at this point in time i think using 24bit for playback is a overreaching a bit. It's sort of like how 64bit operating systems came out too early for 90% of users to have more than 3gb of ram without knowing how to use it in a 32bit OS. In the studio 24bit is definitely the way to go at the moment for the dynamic range, although the noise floor you hear is more likely to come from the equipment than the signal.
**if the OP still disagrees, perhaps he'd like to find and ABX an unshaped and undithered test tone... oh well, i'm done with this before it develops into a flame war...
to save him time he could take it two ways from here
a. disk space to which i have no response but get bigger storage and b. he could argue whether or not differences are audible outside of test tones (i.e. in music). In which case there is no definite answer at the moment. "no" would be the current answer, but it has not really been tested whether it is possible by a large number of people. There are those out there that believe emotional response is more real from higher resolution/sample rates due to the affect of high frequencies on the brain etc.

 

[Arjisme]

Quote:

Originally Posted by CDBacklash /img/forum/go_quote.gif You do know amplitude quantisation occurs in all digital media where each waveform is not represented as a complete picture, but merely as a snapshot, right?

Did you read the first post in this thread?

Quote:

Originally Posted by gregorio /img/forum/go_quote.gif The problem is that an analogue waveform is constantly varying. No matter how many times a second we measure the waveform or how many bits we use to store the measurement, there are always going to be errors. These errors in quantifying the value of a constantly changing waveform are called quantisation errors.

 

[gregorio]

CD Backlash:- You need to read this thread carefully and understand it before you come out with a whole bunch of statements based on assumption which are incorrect. For example:

Quote:

Originally Posted by CDBacklash;5553024 /img/forum/go_quote.gif Sampling rate does not eliminate highs it simply causes aliasing and misrepresentation of them.

Oh please! You can't have "aliasing and misrepresentation" of the recorded waveform otherwise it's going to be a recording full of complete rubbish and no one would be using digital audio. This is why all ADCs have an anti-alias filter, it's not optional, it is an absolute requirement and it's not based on bit depth but purely the Nyquist Point, IE. roughly half the sampling frequency. What college did you go to that taught you the rubbish you are spouting?

Quote:

Originally Posted by CDBacklash /img/forum/go_quote.gif As for your precious 16bit, a lot of studios are now recording in 24bit and then converting 16bit for CD media (oh no!).

That's hardly a suprise to me, as you would know if you'd bothered to read this thread before trolling! I started using high bit rates for recording when you were probably still in diapers so read the damn thread!

Quote:

Originally Posted by CDBacklash /img/forum/go_quote.gif Digital storage mediums will _never_ do audio true justice because of the quantisation that occurs... You do know amplitude quantisation occurs in all digital media where each waveform is not represented as a complete picture, but merely as a snapshot, right? ...

And you do know that a dithering quantiser was invented probably before you were born which converts the interpolation errors during quantisation into un-correlated noise and results in a perfectly linear reproduction of the original waveform ("complete picture"), right? Obviously not!

Quote:

Originally Posted by CDBacklash /img/forum/go_quote.gif There are more ticks to snap to in a 24bit digital recording and it will be a more accurate representation of the sound if done correctly.

Now you are completely getting yourself screwed up. Ticks are the basic timing units derived from the BPM (PPQN) of your sequencer and have nothing to do with sampling rates, bit depth or indeed digital audio! Or maybe you are confusing the quantising of MIDI data with the quantising of digital audio, the same term but totally unrelated processes. If you mean there are more quantisation levels in 24bit then yes there are, as explained in my original post!!!

Quote:

Originally Posted by CDBacklash /img/forum/go_quote.gif Most engineers will apply a brickwall (that is a compression filter with a small attack and release time) and then increase the volume of the track.

You've even got this wrong. I take it you are refering to brickwall compression, which if you knew anything about audio engineering is actually called limiting. The difference between a limiter and a compressor is not the attack and release times but the ratio, which for a limiter needs to be infinity:1. And you've really displayed your ignorance as if you compress or limit the track you are decreasing the dynamic range and therefore require a lower bit depth to encode it. I would call the college you attended and ask for your money back!!

Quote:

Originally Posted by CDBacklash /img/forum/go_quote.gif 24bit audio has the capacity to store orchestra hits or heaven forbid cannon blasts (from a particular epic) far more realisticly because of the dynamic range...

Exactly, so hands up anyone out there that wants to put on headphones and hear the true volume of what a cannon being fired from 3ft away actually sounds like?

To be honest, it's hard to find a single sentence in your post which bares any resemblance to how digital audio actually works. My guess is that you are a kid who has just left college and who has completely mis-heard or mis-understood what your lecturer has tried (but obviously failed) to teach you!

G

 

[CDBacklash]

I did, but my english is poor (and so is my memory).
He makes some valid points however I feel that he is misguided. There is a difference between 16 bit and 24bit audio. Whether current hardware can present it well enough and whether our ears can "get it" well enough is a completely different story.
There was no myth exploded. Everyone knows that it's much easier to ABX 8bit vs 16 bit (or even 12 bit) than it is to abx 16 vs 24. Some of his arguments like that there is "no difference other than dynamic range" are borderline moronic when there clearly are.
exiting stage left
edit: uh no, I'd appreciate if the personal attacks were kept to a minimum.
A compression filter with 1 sample attack and release time does exactly the same thing as a limiter... You dont require a lower bit depth to do anything when you compress something. You can encode anything however the hell you want.
As for the cannon being fired? Yes, I will take the real volume. Exposure time for 120 db is 8.5 minutes if i recall correctly and it is halved for a 5 db increase. A cannonblast amongst a loud orchestra will not cause permenant ear damage.
"Ticks are the basic timing units derived from the BPM (PPQN) of your sequencer and have nothing to do with sampling rates"
I was using tick as an example of where the quantisation occurs (that is vertically on the typical representation of a waveform), not of anything else. This is at best an english error and I would appreciate you to not make fun of my english as it is not my native language.
"Oh please! You can't have "aliasing and misrepresentation" of the recorded waveform otherwise it's going to be a recording full of complete rubbish and no one would be using digital audio. This is why all ADCs have an anti-alias filter, it's not optional, it is an absolute requirement and it's not based on bit depth but purely the Nyquist Point, IE. roughly half the sampling frequency. What college did you go to that taught you the rubbish you are spouting?"
I was talking about in the DAC process, not the ADC. Again i feel like you are making fun of my english... sigh
When you tell me that a digital storage results in a perfect reproduction of the complete picture, you are telling lies. The picture you are hearing is not the original picture. Close, but not quite. Have a nice day, I don't feel like partaking in a flame war. HAF Go here and have a nice day... i'm sure you'll learn a lot.

 

[Clutz]

Quote:

Originally Posted by CDBacklash /img/forum/go_quote.gif Close, but not quite. Have a nice day, I don't feel like partaking in a flame war.

Dude, this is weak. First you accuse him of being a troll- which was both unfounded and uncalled for, and then you say that he's starting a flame war? You threw the first stone. You accused him of not knowing what he was talking about, when the problem was that you couldn't be bothered to read the whole thread. He's been a member for over a year, and you've been a member for less than a month- and yet despite the fact that you started with unnecessary rhetoric, you accuse him of being a troll?

I think you are a troll.

Quote:

This is at best an english error and I would appreciate you to not make fun of my english as it is not my native language.

And stop trying to hide behind your poor English skills. You're out and out attacking people, and then hiding behind your poor English writing skills? That's way weak. Regardless of what your level of English proficiency, we are forced to understand what you mean as a function of what you write. Do you really expect everyone to read your posts and then spend a lot of time analyzing what you meant versus what you wrote? He did not intentionally misinterpret you, that much is clear.

Weak!

 

[CDBacklash]

What does join date have to do with anything? (other than nothing).
I am also aware that I said something about his opinion first, because I was pretty disgusted by how blatantly he was trying to force his opinion (not to mention itd be amazing if he could preempt my post) on people on an undoubtedly grey area (at this point in time). Since then the thread has degraded into tit-for-tat pseudo-flaming which is why I am now ejecting myself. I have said what I wanted to said and am going to leave it at that.
Please keep your personal attacks away from me.

 

[evilking]

Quote:

Originally Posted by CDBacklash /img/forum/go_quote.gif What does join date have to do with anything? (other than nothing). I am also aware that I said something about his opinion first, because I was pretty disgusted by how blatantly he was trying to force his opinion (not to mention itd be amazing if he could preempt my post) on people on an undoubtedly grey area (at this point in time). Since then the thread has degraded into tit-for-tat pseudo-flaming which is why I am now ejecting myself. I have said what I wanted to said and am going to leave it at that. Please keep your personal attacks away from me.

You come in here, spew seven shades of ******** and now you're leaving the thread?

Thanks so much for your input. I'm sorry the language barrier has it made so difficult for you to have a rational debate.


EK

 

[gregorio]

Quote:

Originally Posted by CDBacklash /img/forum/go_quote.gif I was using tick as an example of where the quantisation occurs (that is vertically on the typical representation of a waveform), not of anything else. This is at best an english error and I would appreciate you to not make fun of my english as it is not my native language.

I'm not having a go at your English, I'm having a go at your use of audio engineering terminology. The term "Ticks" is used all over the world including in non-english speaking countries! The correct term for what you are trying to describe is the "sampling point".

Quote:

Originally Posted by CDBacklash /img/forum/go_quote.gif I was talking about in the DAC process, not the ADC. Again i feel like you are making fun of my english... sigh

I'm not falling for your lame excuse that English is not your first language, to try and cover up the fact that you have posted a pile of nonsense. I'm a foreign language speaker too, or didn't you notice my name (Gregorio)? And if you were talking about a DAC and not an ADC then you were even more wrong! It is the ADC which performs the quantisation process, not a DAC. It is an ADC which sets the sampling rate and an ADC which requires an anti-alias filter to remove frequencies above the Nyquist Limit.

Quote:

Originally Posted by CDBacklash /img/forum/go_quote.gif When you tell me that a digital storage results in a perfect reproduction of the complete picture, you are telling lies.

Go away and learn your facts, I'm not going to tell you again! I am not telling you lies, I am directly quoting Harry Nyquist and the Nyquist-Shannon Sampling Theorem:

"In essence the theorem shows that an analog signal that has been sampled can be perfectly reconstructed from the samples if the sampling rate exceeds 2B samples per second, where B is the highest frequency in the original signal." - Nyquist–Shannon sampling theorem - Wikipedia, the free encyclopedia

Maybe you would like to write the word "Liar" on Nyquist's grave stone and while you're at it get a nobel prize for proving the theorem wrong and that digital audio does not exist?

Quote:

Originally Posted by CDBacklash /img/forum/go_quote.gif A compression filter with 1 sample attack and release time does exactly the same thing as a limiter...

And how many professional compressors have you used which have the option to set the attack and release times to 1 sample? How much the signal is compressed is set with the ratio, the attack and release times just set how quickly the compression ratio is achieved and how long it is held once the threashold has been reached. Again the difference between a limiter and a compressor is the ratio, which is variable on a compressor but fixed at infinity:1 on a limiter. Once the track has been compressed the dynamic range has been reduced and therefore can be encoded with a lower bit rate, what part of this simple explanation do you not understand?

Quote:

Originally Posted by CDBacklash /img/forum/go_quote.gif edit: uh no, I'd appreciate if the personal attacks were kept to a minimum.

Could you please go and look up the english word "hypocrite". So far you've called me a moron, a 12 year old and a liar.

Quote:

Originally Posted by CDBacklash /img/forum/go_quote.gif Exposure time for 120 db is 8.5 minutes.

OK, that has dealt with the maximum perceivable dynamic range of a CD. What does it say about 144dB (24bit). I think you'll find it says permanent hearing damage will occur instanteneously. It is also likely that an instantaneous level of 120dB will severely damage the hearing of a child (World Heath Organisation). So you tell me, how irresponsible would a record label have to be to sell a recording which will literally deafen the people who buy it, how long before they got sued out of existance and the directors imprisoned for corporate negligence? Even if you are completely ignorant of the facts, at least try to apply some logic.

For anyone else reading this thread, please do not try and listen to 120dB signals, even for a short time!!

G