Is the dynamic range of CD sufficient?

[microstrip]

Jack, I must be missing something important, because that read like an elaborate version of the old sampling fallacy that sees digital as a series of stairsteps, with lost information between them, and, therefore, the faster the sampling, the lower the loss, the closer to real...

But I know you know that represents a fundamental misunderstanding of sampling and that if it were true, the faster the sampling, the better the output would be. 192 would have to be better than 88. Yeah, I'm confused...

Tim

Tim,

I fail to see any fallacy in what you write - digital is really a series of steps, with lost information between them. If this information is relevant is what people debate. BTW, the output of a DAC, however it is not a series of stairsteps . And theoretically 192 should be be better than 88 - if it is not, it is surely due to the implementation, not the sampling rate. It is why we find interesting debating this preference for 88/96, that has been expressed by several people.

 

[Phelonious Ponk]

Tim,

I fail to see any fallacy in what you write - digital is really a series of steps, with lost information between them. If this information is relevant is what people debate. BTW, the output of a DAC, however it is not a series of stairsteps . And theoretically 192 should be be better than 88 - if it is not, it is surely due to the implementation, not the sampling rate. It is why we find interesting debating this preference for 88/96, that has been expressed by several people.

I know.

Tim

 

[Al M.]

Tim,

I fail to see any fallacy in what you write - digital is really a series of steps, with lost information between them.

Again, this video explains why the stair step idea with 'lost information' is a fallacy:

http://www.youtube.com/watch?v=cIQ9IXSUzuM

 

[JackD201]

Jack, I must be missing something important, because that read like an elaborate version of the old sampling fallacy that sees digital as a series of stairsteps, with lost information between them, and, therefore, the faster the sampling, the lower the loss, the closer to real...

But I know you know that represents a fundamental misunderstanding of sampling and that if it were true, the faster the sampling, the better the output would be. 192 would have to be better than 88. Yeah, I'm confused...

Tim

Hi Tim,

Lots of people get confused about that stair step. I know I did. What it is is just a visual representation and should not be taken literally. What I am talking about is bit depth and how increasing bit depth increases the number of voltages that can be represented between the maximum signal and the minimum signal. It has nothing to do with the sampling rate in a sense because the sampling process happens before the quantization process. Time is converted from continuous to discreet then in quantization the continuous amplitude is converted to discreet. For example a 2bit signal with a max of 1 and a minimum of -1 would yield 4 voltage representations separated by .5 each. Add bits and that .5 shrinks mighty quickly. The sampling frequency determines the frequency range. So 2bits at 4 values can have a high of ~22kHz at 44.1 and double that at 88.2, both will still have intervals of .5 between highest and lowest when it comes too amplitude regardless of frequency.

 

[Phelonious Ponk]

Hi Tim,

Lots of people get confused about that stair step. I know I did. What it is is just a visual representation and should not be taken literally. What I am talking about is bit depth and how increasing bit depth increases the number of voltages that can be represented between the maximum signal and the minimum signal. It has nothing to do with the sampling rate in a sense because the sampling process happens before the quantization process. Time is converted from continuous to discreet then in quantization the continuous amplitude is converted to discreet. For example a 2bit signal with a max of 1 and a minimum of -1 would yield 4 voltage representations separated by .5 each. Add bits and that .5 shrinks mighty quickly. The sampling frequency determines the frequency range. So 2bits at 4 values can have a high of ~22kHz at 44.1 and double that at 88.2, both will still have intervals of .5 between highest and lowest when it comes too amplitude regardless of frequency.

Thanks, Jack. With that, I think I'll let this go before it devolves into another D v A thread.

Tim

 

[Al M.]

Thanks, Jack. With that, I think I'll let this go before it devolves into another D v A thread.

Tim

Tim, I like your signature

 

[GaryProtein]

After reading all the posts, I am still wondering, is there not enough dynamic range to capture the loudest sounds we need in a recording or not enough dynamic range to capture the softest sounds in a recording?

I have a birth fetish and would like to record the sound of a tubercle bacillus dividing in the lung of a jackhammer operator in a construction site.

 

[JackD201]

Thanks, Jack. With that, I think I'll let this go before it devolves into another D v A thread.

Tim

Okidoki Tim. I wish I could have found a better example of bit depth than this one but I hope it helps in illustrating the differences with the same frequency content, Gotta run too.

You can hear the difference in overtones even in compressed form (like the film watched on TV example). Imagine what that might sound like if we were there.

 

[Julf]

You can hear the difference in overtones even in compressed form (like the film watched on TV example). Imagine what that might sound like if we were there.

So I think we can all agree that 8 bits is not quite enough to reproduce "normal" sounds (it might still work for some optimized digital sound effects).

 

[Phelonious Ponk]

After reading all the posts, I am still wondering, is there not enough dynamic range to capture the loudest sounds we need in a recording or not enough dynamic range to capture the softest sounds in a recording?

I have a birth fetish and would like to record the sound of a tubercle bacillus dividing in the lung of a jackhammer operator in a construction site.

I believe we have a new standard.

Tim

 

[microstrip]

Again, this video explains why the stair step idea with 'lost information' is a fallacy:

http://www.youtube.com/watch?v=cIQ9IXSUzuM

Al. M.

Sorry I have not the interest or time to see or comment videos on this elementary subject. If you want to express an argument or idea in words I will be pleased to answer or debate it.

 

[Al M.]

Al. M.

Sorry I have not the interest or time to see or comment videos on this elementary subject. If you want to express an argument or idea in words I will be pleased to answer or debate it.

Hi Microstrip,

I could not express it any clearer than the video vividly demonstrates, and it would take me more time to come up with a written explanation -- which would still be less powerful than the video -- than it would take you time to actually watch the video. If you have no interest to watch, there is nothing to debate, I am afraid.

 

[microstrip]

Hi Microstrip,

I could not express it any clearer than the video vividly demonstrates, and it would take me more time to come up with a written explanation -- which would still be less powerful than the video -- than it would take you time to actually watch the video. If you have no interest to watch, there is nothing to debate, I am afraid.

Al. M,

My problem is understanding what you mean by it. A digital process intorduces steps and looses information - due to quantization errors and sampling.
I can not understand why a "powerful video" lasting for 24 minutes is needed to debate simple thinks. My fault.

 

[Al M.]

Al. M,

A digital process intorduces steps and looses information - due to quantization errors and sampling.

Again, explained in the video (shrug).

By the way, you would only need to watch the first 10 minutes or so.

 

[microstrip]

Again, explained in the video (shrug).

By the way, you would only need to watch the first 10 minutes or so.

To find what?

 

[esldude]

To find what?

To find out thinking in stairsteps is wrong and that nothing is lost between samples even compared to all generation and monitoring with quality instruments in analog. You also will find out what Shannon Nyquist really means.

 

[microstrip]

To find out thinking in stairsteps is wrong and that nothing is lost between samples even compared to all generation and monitoring with quality instruments in analog. You also will find out what Shannon Nyquist really means.

Esldude,

Any digital process has steps and the process of removing these steps is not innocuous. Also something is lost between samples, weather these losses are relevant to digital sound reproduction is IMHO what is being debated. And I can assure you I know what Shannon Nyquist means since long ... Unfortunately when time and amplitude get mixed things are not so easy. The debates on the Trinity DAC were a clear proof of that.

 

[GaryProtein]

To find what?

It is really worthwhile to view the video.

You will get more information than you expect.

 

[esldude]

Esldude,

Any digital process has steps and the process of removing these steps is not innocuous. Also something is lost between samples, weather these losses are relevant to digital sound reproduction is IMHO what is being debated. And I can assure you I know what Shannon Nyquist means since long ... Unfortunately when time and amplitude get mixed things are not so easy. The debates on the Trinity DAC were a clear proof of that.

Yeah, just watch the video. Please. You can argue all you want. The only thing missed between sample points will be something too high in frequency to hear. If low enough in frequency to hear it will be picked up and reconstructed properly both in amplitude and time. When you can generate clean analog signals, monitor the result of an DA to AD conversion with good analog spectrum analyzers and scopes seeing you lost nothing then what is the problem? If you view the video, and do understand Shannon-Nyquist then the only debate left in regard to digital sound reproduction is bandwidth.

In the time you have been mucking about regarding what is in the video you could have watched it two or more times already.

 

[microstrip]

Yeah, just watch the video. Please. You can argue all you want. The only thing missed between sample points will be something too high in frequency to hear. If low enough in frequency to hear it will be picked up and reconstructed properly both in amplitude and time. When you can generate clean analog signals, monitor the result of an DA to AD conversion with good analog spectrum analyzers and scopes seeing you lost nothing then what is the problem? If you view the video, and do understand Shannon-Nyquist then the only debate left in regard to digital sound reproduction is bandwidth.

In the time you have been mucking about regarding what is in the video you could have watched it two or more times already.

No further comments are needed. Have a good day.