[Davesrose]

Man, every time I check this post, it derails to something else! This time oversampling. To get back to an analogy that was used for digital photos.....resampling to a higher resolution is not as good as actually recording at that resolution. The best that happens when you resize a photo to a higher color depth or resolution, is that the image may become less aliased with dithering. But it's still no where as vibrant or sharp as an image that's recorded in that native resolution/color depth. That's why in the graphics world, source images are always captured at a larger resolution then the final output: you don't get nearly as many distortions resizing an image down as you would scaling it up.

For the points about audio sampling.....yes, equipment that oversamples can have an audible difference. It certainly colors the original 16 bit source: whether it's good or not is up to the listener's interpretation. On my setup, I've noticed a bigger change in sound difference going from redbook to SACD: not regular rebook vs oversampled redbook. It's a similar analogy as upconverting DVD players. A regular DVD upconverted to 1080i seems to have a faint difference with a 480p image: but there's a much larger difference with an image that was actually taken at 1080p.

 

[hciman77]

Quote:

Originally Posted by drarthurwells /img/forum/go_quote.gif What you should say about my facts is that they disturb you, and you refuse to face their reality, and thus you will continue to deny they are real and will criticize me for pointing out the truth. Most people do that much of the time.

I would watch the Messiah complex Art, could get you into trouble, plus it makes you sound a lot like Andrea

 

[tourmaline]

Quote:

Originally Posted by hciman77 /img/forum/go_quote.gif Er a Byte is 8 bits, a bit can be 0 or 1 and a byte will have 8 bits all of which will be 0 or 1, unless there is a sign bit. So for instance 16 bits gives you a number from 0 - 65535 but with a sign bit you get from - 32767 to + 32767.

True, but i mean 0 is no info 1 is audio info....the more 1's are used in a bit the more info you have...zero is no info thus aloose bit in this case...in digital it is seen as a zero but it contains no audible info...

eg i bit with 11111111 contains more technical info then a bit that contains this: 00000000...in this case it means it's nothing.....
technically you can use 16 bit but practically only 15 bits or less...if the last 2 bits are all zero's, technically you end up with 14 bits...

 

[jmmtn4aj]

I'm not even going to attempt joining the fray

For me, the attraction of vinyl is the imperfections of an analogue medium. I love the sound of the needle landing on the surface, and the clicks and pops throughout the tracks. It's sort of like the whole analogue versus digital thing going on in cinemas. For me, the experience comes from the imperfections in the audio and film strip that makes watching a movie in the cinema feel like watching a movie in the cinema.

 

[hciman77]

Quote:

Originally Posted by tourmaline /img/forum/go_quote.gif True, but i mean 0 is no info 1 is audio info....the more 1's are used in a bit the more info you have...zero is no info thus aloose bit in this case...in digital it is seen as a zero but it contains no audible info...

Er, no , information is information, the 16bits gives you level ranges of 0 - 65535 i.e the dynamic range of 96db an individual bit being zero is just part of the number so for instance

1111111111111111 =
1 + 2 + 4 + 8 + 16 + 32 + 64 + 128
+ 256 + 512 + 1024 + 2048 + 4096 + 8192 + 16384 + 32768 = 65535 or
the highest level possible

While 1111011111111111 =
1 + 2 + 4 + 8 + 16 + 32 + 64 + 128
+ 256 + 512 + 1024 + 0 + 4096 + 8192 + 16384 + 32768 = 63487
which is slightly lower

To get a much lower level you have to make the most significant digit zero, which halves the possible level (give or take).

 

[Leporello]

Quote:

Originally Posted by tourmaline /img/forum/go_quote.gif As you know bits can store 1's and 0's...more 0's is less info...more 1's is more info and detail...

Sorry, but this is nonsense.


Regards,

L.

 

[hciman77]

Quote:

Originally Posted by tourmaline /img/forum/go_quote.gif eg i bit with 11111111 contains more technical info then a bit that contains this: 00000000...in this case it means it's nothing..... technically you can use 16 bit but practically only 15 bits or less...if the last 2 bits are all zero's, technically you end up with 14 bits...

No, you have 16 bits and always have 16 bits regardless of the contents of those bits, that is why digital silence is really quiet it is 0000000000000000 i.e no signal at all. The bits give you a signal level, this is roughly equivalent to the pattern of the groove in vinyl which when the needle tracks it generates more or less vibration in the cantilever which causes an electrical signal of a greater or less magnitude to be generated by the magnet moving in an electric field. You wouldnt argue that just because the magner doesnt move much for quiet signals that you are losing resolution.

If the most significant two digits are 0 then you just have a quieter signal and in classical music this is often what you want.

 

[hciman77]

Quote:

Originally Posted by Leporello /img/forum/go_quote.gif Sorry, but this is nonsense. Regards, L.

True, but you could explain why.

 

[Febs]

Quote:

Originally Posted by drarthurwells /img/forum/go_quote.gif What you should say about my facts is that they disturb you,

You're right. It disturbs me that what you cite as "facts" do not appear to have any empirical support.

Quote:

and you refuse to face their reality, and thus you will continue to deny they are real

Again, you're right. I will continue to deny the reality of your "facts" until you show me that they are, in fact, real. You haven't yet done so, and you don't appear to be able to do so.

Quote:

and will criticize me for pointing out the truth.

No, I criticize you for relying on statistics that do not appear to have any factual basis.

 

[tourmaline]

Quote:

Originally Posted by hciman77 /img/forum/go_quote.gif No, you have 16 bits and always have 16 bits regardless of the contents of those bits, that is why digital silence is really quiet it is 0000000000000000 i.e no signal at all. The bits give you a signal level, this is roughly equivalent to the pattern of the groove in vinyl which when the needle tracks it generates more or less vibration in the cantilever which causes an electrical signal of a greater or less magnitude to be generated by the magnet moving in an electric field. You wouldnt argue that just because the magner doesnt move much for quiet signals that you are losing resolution. If the most significant two digits are 0 then you just have a quieter signal and in classical music this is often what you want.

Ok,this might be true for audio but for data it is valid...you won't get any responce...
I know 16 bits is 16 bits...it is filled with zero's....

So, if those 2 bits are noise then you only have 14 bits with true audio information, right. That i was trying to tell you...and the other two bits generate static noise. but the fact is that you use only 14 bits of info out of 16...by using or not using all bits you can alter the sound...(background noise).

If not clear me up.

 

[tourmaline]

Quote:

Originally Posted by Febs /img/forum/go_quote.gif You're right. It disturbs me that what you cite as "facts" do not appear to have any empirical support. Again, you're right. I will continue to deny the reality of your "facts" until you show me that they are, in fact, real. You haven't yet done so, and you don't appear to be able to do so. No, I criticize you for relying on statistics that do not appear to have any factual basis.

Neither have you, you only scream not true or nonsense...you need to fill us in...we can learn from both sides.

 

[hciman77]

Quote:

Originally Posted by tourmaline /img/forum/go_quote.gif So, if those 2 bits are noise then you only have 14 bits with true audio information, right. That i was trying to tell you...and the other two bits generate static noise. but the fact is that you use only 14 bits of info out of 16...by using or not using all bits you can alter the sound...(background noise). If not clear me up.

A zero is a zero not noise. Noise is when you have stuff in the replay that is not meant to be there i.e hum or harmonic distortion or some error, when the high order bits are zero and are meant to be zero then all is well, if a bit should be zero and is misread as 1 (or vide versa) then that is error or noise or distortion call it what you will.

You have 16 bits and each will be zero or 1, the pattern of 0s and 1s tells you the level of the signal, if you never have a signal that sets the most significant bit to 1 then you just have a lower dynamic range on the recording, it is like having a quiet record it doesnt meant that the lp playback is worse just that the recording is not using the full capability of the medium.

 

[Febs]

Quote:

Originally Posted by tourmaline /img/forum/go_quote.gif Neither have you, you only scream not true or nonsense...you need to fill us in...we can learn from both sides.

tourmaline, you yourself said this just a few posts ago: "A partial answer is provided by the Nyquist–Shannon sampling theorem, which provides a sufficient (but not always necessary) condition under which perfect reconstruction is possible. The sampling theorem guarantees that bandlimited signals (i.e., signals which have a maximum frequency) can be reconstructed perfectly from their sampled version, if the sampling rate is more than twice the maximum frequency. Reconstruction in this case can be achieved using the Whittaker–Shannon interpolation formula." (The emphasis is mine.)

What drarthurwells is saying appears to be a direct challenge to the Nyquist theorem, yet he adamantly refuses to provide any support for that challenge. If his statement is, in fact, true, then he should be able to provide some support for it. Instead, he simply points to the statement of Julian Hirsch, which is unsupported by any sort of evidence whatsoever. With all due respect for Mr. Hirsch, we do not need to accept the assertion that "The ones and zeros of digital are only 5 per cent of the ones and zeros theoretically possible in the whole original" just because he or drarthurwells say so. If drarthurwells wants anyone to believe his "5%" assertion, then he should at the very least be able to explain how the 5% figure is derived. (For example, why is it 5%, and not 1%, or 20%, or 50%?)

 

[drarthurwells]

Art: What you should say about my facts is that they disturb you,

Febs: It disturbs me that what you cite as "facts" do not appear to have any empirical support.

Art: ...and you refuse to face their reality, and thus you will continue to deny they are real.

Febs: Again, you're right. I will continue to deny the reality of your "facts" until you show me that they are, in fact, real. You haven't yet done so, and you don't appear to be able to do so.

Art: ....and will criticize me for pointing out the truth.

Febs: No, I criticize you for relying on statistics that do not appear to have any factual basis.

Quote:

Originally Posted by Febs /img/forum/go_quote.gif tourmaline, you yourself said this just a few posts ago: "A partial answer is provided by the Nyquist–Shannon sampling theorem, which provides a sufficient (but not always necessary) condition under which perfect reconstruction is possible. The sampling theorem guarantees that bandlimited signals (i.e., signals which have a maximum frequency) can be reconstructed perfectly from their sampled version, if the sampling rate is more than twice the maximum frequency. Reconstruction in this case can be achieved using the Whittaker–Shannon interpolation formula." (The emphasis is mine.) What drarthurwells is saying appears to be a direct challenge to the Nyquist theorem, yet he adamantly refuses to provide any support for that challenge. If his statement is, in fact, true, then he should be able to provide some support for it. Instead, he simply points to the statement of Julian Hirsch, which is unsupported by any sort of evidence whatsoever. With all due respect for Mr. Hirsch, we do not need to accept the assertion that "The ones and zeros of digital are only 5 per cent of the ones and zeros theoretically possible in the whole original" just because he or drarthurwells say so. If drarthurwells wants anyone to believe his "5%" assertion, then he should at the very least be able to explain how the 5% figure is derived. (For example, why is it 5%, and not 1%, or 20%, or 50%?)

Art: So, we should believe you instead of Julian Hirsch, editor of Hi Fi and Stereo Review (later just Stereo Reveiw) magazine - a foremost authority in music reproduction in the 40s, 50s, 60s, 70s, 80s, etc.?

Where is anyone's proof to refute my report of what Hirsch said (at some point in the 70s, not sure of the magazine issue date)?

You have not one shred of information to refute the fact that digital recording is a 5 per cent sample of original musical data as I report that the Hirsch article stated in talking about the then new digital recording technology. You do not know. Yet you have no problem contradicting what you simply do not know - incredibile! The sad fact is that the world is full of people like you - hanging onto false misconceptions which they refuse to change in spite of any contradiction.

Think of a curve as an analog waveform of musical data. There are many data points connected together to make up this curve, strung together like electrons in orbit around a nucleus of an atom. Now digital recording samples these data points in consistent regularity (timing) at a rate of one every twenty. A dotted curve from this digital sample is unconnected dots instead of a line curve, with a space of 19 missing dots between each dot represented on the lined curve. In listening, the brain connects the dots and perceives the data as a complete curve. However, this digital curve is weaker in some ways than the analog curve in the same way that a hologram reproduced from a 5 per cent sample of the original is not perfectly identical to the original.

Instead of saying and moaning, "That can't be true!", someone refute my assertions with a credible source other than, "I don't like that reality so I will refuse to believe it, and criticize anyone who states it, in spite of any contradictory evidence."

I believe what Julian Hirsch published. I never saw anyone disagree with him in any subsequent publications, have never seen any evidence to the contrary from any source other than on this forum from people who simply don't like the reality so refuse to believe it, and criticize anyone who states it, in spite of any contradictory evidence.

 

[tourmaline]

Quote:

Originally Posted by Febs /img/forum/go_quote.gif tourmaline, you yourself said this just a few posts ago: "A partial answer is provided by the Nyquist–Shannon sampling theorem, which provides a sufficient (but not always necessary) condition under which perfect reconstruction is possible. The sampling theorem guarantees that bandlimited signals (i.e., signals which have a maximum frequency) can be reconstructed perfectly from their sampled version, if the sampling rate is more than twice the maximum frequency. Reconstruction in this case can be achieved using the Whittaker–Shannon interpolation formula." (The emphasis is mine.) What drarthurwells is saying appears to be a direct challenge to the Nyquist theorem, yet he adamantly refuses to provide any support for that challenge. If his statement is, in fact, true, then he should be able to provide some support for it. Instead, he simply points to the statement of Julian Hirsch, which is unsupported by any sort of evidence whatsoever. With all due respect for Mr. Hirsch, we do not need to accept the assertion that "The ones and zeros of digital are only 5 per cent of the ones and zeros theoretically possible in the whole original" just because he or drarthurwells say so. If drarthurwells wants anyone to believe his "5%" assertion, then he should at the very least be able to explain how the 5% figure is derived. (For example, why is it 5%, and not 1%, or 20%, or 50%?)

I agree with ya on this one...

If i read the info correct...it has to be way more then 5%, even a perfect copy of the original sinus is possible...in short, theorectically it is possible to fully copy the original sound, if the condition is right, wich has to be double the sampling rate...so, if you have sound of 21 khz, you need a sampling rate that is at least 42...i see no problem there, maybe a brickwall filter or soemthing kicks in...but in theory, it is possible to sample this sound without any problem...if you take the sampling rate to 24 bits and 192 khz then i don't see any problem to get all the sounds available without any limitation...

even a sound going to almost 100khz is covered with this method...so in real world this would be enough to sample the complete sound without any distortion or limits...

The figure of 5% still doesn't look really viable to me...

We all know by now that the recording gear is able to record everything in the sound...the only limitation is the 16 bits 44.1 khz cd...even then, according to the above written theorem, we still have a full range of at least 22 khz...so, we are able to extract everything in the recording that is possible. Recording studio's now use even 64 bit recording or processing... why not move to 32 bits?

it's also a fact that barely people can hear to 20khz...