[b0dhi]

You're really making my night here, gregorio. I haven't laughed so hard in a while.

Thanks for that unrelated quote on dithering from Wikipedia. I'll repeat myself for the third time: -

Quote:

Originally Posted by b0dhi /img/forum/go_quote.gif Hahahahahahahahahahaahaha. Sorry that is just hilarious. I'm well capable of understanding your "explanation", I just happen to know it's incorrect, which is why I'm asking you to supply a reference, because I know you won't be able to. Please - since you're a university lecturer who is aware of what the Harvard referencing system is (lol), you should have no problem finding some lecture notes stating that "the resolution of the audio is just as perfect [equal] at 16bit as it is at 24bit or indeed at any number of bits". Although I'm not sure how you will do that due to the fact that time-domain resolution depends directly on bit-depth (given enough distance from Fs/2), and you might also have some trouble convincing me that a 1-bit dithered stream has the same "resolution" (that word, by the way, you are confusing because it relates more to sampling rate than bit-depth, but I will assume you're talking about quantisation resolution) as a 16-bit dithered stream.

But if you want to quote Wikipedia, then have a squizz at this:
Quote:

Bit depth directly corresponds to the resolution of each sample in a set of digital audio data.

(from Audio bit depth - Wikipedia, the free encyclopedia)

 

[nick_charles]

Quote:

Originally Posted by b0dhi /img/forum/go_quote.gif But if you want to quote Wikipedia, then have a squizz at this: (from Audio bit depth - Wikipedia, the free encyclopedia)

I would hesitate to use Wikipedia as anything other than a starting point for the sensible references, as for your wikipedia quotation it is prefaced by the following warning.

The current version of the article or section is written like a magazine article; it does not use the direct, balanced tone expected of an encyclopedia.

i.e it is in the opinion of Wikipedia (and their high standards) a fluff piece

Anyway , what evidence do you have for suggesting that 16 bits is inadequate ?. Can you point to some proper listening tests ?

 

[b0dhi]

Quote:

Originally Posted by nick_charles /img/forum/go_quote.gif I would hesitate to use Wikipedia as anything other than a starting point for the sensible references, as for your wikipedia quotation it is prefaced by the following warning. The current version of the article or section is written like a magazine article; it does not use the direct, balanced tone expected of an encyclopedia. i.e it is in the opinion of Wikipedia (and their high standards) a fluff piece Anyway , what evidence do you have for suggesting that 16 bits is inadequate ?. Can you point to some proper listening tests ?

The note talks about the tone of the article, in other words, the writing style, not the content. Also, I'm not quoting Wikipedia because I'm trying to prove anything. I just would'nt want gregorio to think he's taken a single step towards supporting his assertion by quoting an unrelated Wiki.

Secondly, I never said that 16bit is not enough. I said I'd like gregorio to support his assertion (quoted previously in bold) with a reference, which he has very amusingly failed to do. The burden of proof is on the claimant, which is gregorio.

 

[gregorio]

b0hdi - The burden of proof is for you to disproove digital audio theory. Glad you had a good laugh, let me guess, was it an idiotic laugh? If you think dithering has got nothing to do with the quality of digital audio go for it. In the meantime here is another explanation:

Article Preview - Digital Problems, Practical Solutions

According to you bit depth and resolution are the same thing. So according to your theory the one bit digital DSD system of SACD must be massively inferior to the 16bit found on CD. Here is a bit of text taken from the SOS forum and written by Hugh Robjohns, the technical editor of Sound on Sound which may help your ignorance b0hdi.

Quote:
my brain refuse to soak in that a 16-bit capture of a waveform has the same 'resolution' as a 24-bit capture.

It because of years of misunderstanding and the wrong terminology. Resolution is a misleading term. The resolution of a linear system is perfect, inherently. The limiting factor is noise. Digital systems have been explained as using quantisation (which they do) and so everyone has this concept of a stepped transfer curve. However, a correctly dithered quantiser doesn't have a stepped transfer curve. It is linear. That's the whole point!

Quote:
If I reduced bit-depth to a 1-bit recording a sine wave would look like a square wave, therefore the resolution of the recording would be poor...

It would only look line a square wave if you were using an undithered quantiser. It would also make SACD sound very poor indeed

Quote:
So, surely a 24-bit recording must be a higher resolution than a 16-bit. Doh, my head...

No. Exactly the same resolution, just a much lower noise floor.

Quote:
What is my brain missing (apart from a few cells)?!

The concept of dither.

I hope d0hdi you are as good at apologising as you are at insulting. My guess is though that you are incapable of admitting you're wrong, that's if you even have the ability to recognise that you're ignorant and making a fool of yourself.

 

[nick_charles]

But if your original waveform has a very high dynamic range of say 105db a 16 bit system would compress it down to 96db and a 8 bit system would compress it even more. Surely this changes the recreated waveform as some information would be lost, the recreated wave may have the same fundamental shape but it would not be the same ?

 

[gregorio]

Nick - No a 16bit system would not compress the waveform, it would reproduce the waveform perfectly. The problem, in your example, is you might not be able to hear the very quietest bits above the noise floor. I say might not because with modern noise shaped dither you can actually hear material below the noise floor. If we go down to 8bits then the waveform is still going to be perfectly linear but our noise floor is going to be even higher and we are definitely not going to be able to hear the quietest parts of the waveform above the noise floor. That's the theory, in practice it is borderline that even in a world class recording studio you would be able to hear a 105dB range, the noise floor of the equipment, the recording venue and the monitoring environment is just too high.

I don't think we should bother about b0hdi too much more Nick, unless he decides to be sensible. He quoted me a wiki article which actually supports my argument but he obviously isn't bright enough to realise it. If you read down the article he quoted from wiki it states: "In pulse-code modulation (PCM) sampling, the bit depth will limit qualities such as dynamic range and signal-to-noise ratio."

 

[shinew]

I just ran across this article and would like to hear people's opinion on this -> Exploring Digital Audio Myths and Reality Part 1 — Audioholics Home Theater Reviews and News

While most of the article is dedicated to explain that at 44.1kHz, an audio signal within human's hearing range can be "perfectly" represented as long as it's a sine wave. It also states that upsampling to higher 96kHz or 192kHz could preserve amplitude and harmonic accuracy of non-sine waves better due to better filtering, such as the sawtooth & square wave which do exist in music.

any thought?

EDIT: by re-reading the last paragraph of the article, i just realized that he actually wrote " a higher sampling rate can help", not upsampling. I assume that they're quite different things.

 

[b0dhi]

Gregorio, so far you've:

- Claimed you're a university lecturer but is unaware of what burden of proof is.

- Referenced an untirely unrelated Wikipedia article (on dithering) to try to support your claim, with a straight face.

- Used ad-hominems, put words in my mouth, and made up fictional quotes.

- You believe that the following two sentences are the same:
"In pulse-code modulation (PCM) sampling, the bit depth will limit qualities such as dynamic range and signal-to-noise ratio." (from Wikipedia)

"the resolution of the audio is just as perfect [equal] at 16bit as it is at 24bit or indeed at any number of bits" (from gregorio)

- Haven't bothered responding to the following quote which directly contradicts your statement:
"Bit depth directly corresponds to the resolution of each sample in a set of digital audio data." (from Wikipedia)


In short, you've done everything but the one thing I asked for, which
was for you to supply a reference supporting your claim that "the resolution of the audio is just as perfect [equal] at 16bit as it is at 24bit or indeed at any number of bits". I'll re-iterate at this point that I assume you're talking about time-domain resolution and/or quantisation resolution, just so we're clear.

 

[nick_charles]

Quote:

Originally Posted by shinew /img/forum/go_quote.gif I just ran across this article and would like to hear people's opinion on this -> Exploring Digital Audio Myths and Reality Part 1 — Audioholics Home Theater Reviews and News While most of the article is dedicated to explain that at 44.1kHz, an audio signal within human's hearing range can be "perfectly" represented as long as it's a sine wave. It also states that upsampling to higher 96kHz or 192kHz could preserve amplitude and harmonic accuracy of non-sine waves better due to better filtering, such as the sawtooth & square wave which do exist in music. any thought? EDIT: by re-reading the last paragraph of the article, i just realized that he actually wrote " a higher sampling rate can help", not upsampling. I assume that they're quite different things.

Christine Tham is a well-known vinyl apologist and has done a few articles critiquing digital and conveniently ignoring vinyl's limitations. Nevertheless this article is more or less ok. However humans cannot tell the difference between a 10K sine wave and a 10k square wave or sawtooth wave so the point is moot...also guess what a square wave is, it is an infinite series of , yes you guessed sine waves...

 

[gregorio]

b0hdi - I said I would respond if you came out with with a sensible question. If you don't know how a dithering quantizer relates to digital audio, then read my posts or do some of your own research. I've already explained why resolution in not an accurate term to use when discussing bit depth. If you want to quote Wiki then go right ahead, I would rather quote the most respected scientist in the field from one of the most respected professional journals.

Now grow up b0hdi, provide some referenced proof that a dithering quantiser doesn't do what I'm suggesting.

Shinew - When upsampling (say from CD), you are taking digital audio which already contains any artefacts caused by 44.1k anti-alias filters. Recording at 96k is different because we are talking about bypassing the 44.1k filters entirely, throughout the recording and playback chain.

 

[gyrodec]

b0dhi

When gregorio hade his number of bits just means noise floor not resolution comment on Friday I knew he was right, but it still sounded crazy to me. He was right but I didn't understand why. (It was helped by the fact the he doesn't try to explain what he says, he just requotes himself. It also didn't help that he insults you for not getting it when he has made no real effort to try to find a way to make you understand - sorry gregorio, but its true).

This morning I woke up and I just understood clearly why he was correct. No idea why it came to me, but here it is.

If we say the signal we are trying to record is at a level such that the MSB is equal to 2v (so all bits on would be ahair under 4v). Adding more bits does NOT made more division between 2v ond 0v (i.e more resolution). In a AD every added bit equates to a voltage half the preceeding one. So 16 bits letes you "see" signals upto 96bd below full volatge. 24 bits let you "see" signals as small as 144 Db below max. BUT, say a volatge at -60dB will be defined by exactly the same digital data on a 16 bit or 24 bit AD, the 24 bit one will simply have more details of the noise floor below the signal. So, gregorio is right, for any givien signal the number of bits doesn't give more detail about the signal, just more room betwen the signal and the noise floor. The 1 bit case pushes this to the limit and the signal is completely burried in the noise floor (mega-oversampling excluded for simplicity), but in theory the signal is in there.

 

[shinew]

not to be a thread police but can we keep the 24bit vs 16bit discussion in the "bit thread" and leave this thread just for kHz discussion?
I think they're starting to merge already. It can confuse people who have not read both of the threads thoroughly(like i did

).

 

[gregorio]

The bit depth thing can get very complicated. I wouldn't have explained it quite the way you did but you've got the basic facts right. Just to bring you up to date of the current technology, because of noise shaped dither we can actually extend the noise floor below the theoretical maximum. The bottom line is that the use of 24bit for the consumer is clear cut.

Unfortunately, the use of higher sample frequencies is not so clear cut. There is growing evidence that 192kFs/s will give a poorer result than 96k, even in recording, let alone upsampling. But at 96k the situation is largely dependent on your DAC.

 

[emmodad]

Quote:

Originally Posted by gregorio /img/forum/go_quote.gif [ snip ] There is growing evidence that 192kFs/s will give a poorer result than 96k, even in recording [ snip ]

An interestingly-worded assertion....but perhaps the word selection is simply sparse. Beyond hardware limitations having to do with performance @ 192 of some current A/D/A integrated circuit implementations (and their architectures) themselves, or current state of capability in associated system hardware implementation, is your wording meant as reference to other problems? Some clarification of your intended meaning could be helpful to folk on this board who don't have technical a/o engineering background in dsp or sampling theory.

 

[nick_charles]

Quote:

Originally Posted by gregorio /img/forum/go_quote.gif Unfortunately, the use of higher sample frequencies is not so clear cut. There is growing evidence that 192kFs/s will give a poorer result than 96k, even in recording, let alone upsampling. But at 96k the situation is largely dependent on your DAC.

Quote:

Originally Posted by emmodad /img/forum/go_quote.gif An interestingly-worded assertion....but perhaps the word selection is simply sparse. Beyond hardware limitations having to do with performance @ 192 of some current A/D/A integrated circuit implementations (and their architectures) themselves, or current state of capability in associated system hardware implementation, is your wording meant as reference to other problems? Some clarification of your intended meaning could be helpful to folk on this board who don't have technical a/o engineering background in dsp or sampling theory.

Dan Lavry did a very interesting White paper about sampling. He discusses the pros and cons of very high sampling rates. Very readable it is too

http://www.lavryengineering.com/docu...ing_Theory.pdf

here is a snippet

The above suggests that 88.2 or 96KHz would be overkill. In fact all the
objections regarding audio sampling at 44.1KHz, (including the arguments relating to preringing of an FIR filter) are long gone by increasing sampling to about 60KHz.

I am not saying this is the word of [Insert preferred deity here] but he does argue the case quite well.