Who here can hear above 22kHz......!!!!????...

[djsubject]

my dog can

[birrbits]

....and if you can't whats the frikking point of 96kHz sampling?!?!!!?

That the bandwidth for nonlinear audio processing is doubled and with proper downsampling from 96 KHz to 48KHz (or 44.1) you get rid of some aliasing artefacts that otherwise would be present in your audible frequency range of sound.

only if you have a cheap analog filter to begin with.

[shamann]

It's not just about capturing frequencies above 22KHz, it's about improving the resolution of frequencies below.

So, at 44.1 KHz sampling rate, you only get two coordinates to ascertain that the recorded frequency is 22.x KHz, four coordinates to ascertain 11.x KHz, etc. With that little information, there is going to be sonic detail that gets lost.

Not essential, as things sound just fine at 44.1 KHz, but finer resolution can't hurt.

[birrbits]

It's not just about capturing frequencies above 22KHz, it's about improving the resolution of frequencies below.

So, at 44.1 KHz sampling rate, you only get two coordinates to ascertain that the recorded frequency is 22.x KHz, four coordinates to ascertain 11.x KHz, etc. With that little information, there is going to be sonic detail that gets lost.

Not essential, as things sound just fine at 44.1 KHz, but finer resolution can't hurt.

My point is that the math proves that it doesn't get lost, that that 'little' amount of information is all that is required to reproduce the analog signal with complete (100%) accuracy. I am not sure who told you about these coordinates, but some dsp reading might help you to understand sampling theory better.

Finer resolutions may not hurt, but they don't add to anything that was 'missing' from our hearing range either. The marketing brag about 96k is a nice way to avoid the issues of spending say $10/unit on a good pre-filter with a 50 cent ADC versus spending $2/unit on the filter and using a $1/unit ADC. Oh and doubling the amount of space eaten up on your hardrive which is cheap anyway.

rant complete

[rockstar_not]

A good text is Proakis and Manolakis: Digital Signal Processing.
Click Here to buy or get the ISBN This was the textbook used in the EE-638 DSP class I went through back in the day at Purdue University.

Does a nice job explaining how sampling with the right D/A conversion does accurately represent content up to the nyquist frequency. Proper anti-aliasing filtering needed at the A/D stage and the right filter response on the D/A stage - but it can be done.

Why are you worrying anyways: Someone will download your music to mp3 or some other crap compression codec which will smear your music way more than the difference between a 16/44.1 and a 24/96 recording!!!

Nobody here is going to get published on SACD or DVD-Audio, unless, of course, BT is lurking here!!


Benefits of 24 bit quantization do exist in a much lower quantization signal to noise ratio, but in order to make use of that, your entire signal chain; analog portions included, have to be dead quiet and your mix engine must mix at 24 or higher bit depth.

-Scott

[birrbits]

I could not agree more about the benefits of 24 bit recording.
but you shouldnt say anything about BT, or my thread might get erased

Incidently, I ta'd out of that book for a semester or 2, its definately one of the better books out there.

[birrbits]

sorry to rant all, please continue to share your 'golden ear' Meffy stories

While I'm at it I'll share mine, one of my projects in grad school was a nice lil audio watermarker, something that you could encode a bunch of text with (to say play back lyrics in real-time ).

Anyway to make a long story short I 'hid' most of the signal above 16kHz. Out of 20 or so guinea pings, one girl claimed to be able to hear it, and described a pinching sensation in her nose.

If you are really bored you can read about it here
http://www.ecel.ufl.edu/~jbirr/watermarking/index.html

[nuffink]

It's not just about capturing frequencies above 22KHz, it's about improving the resolution of frequencies below.

So, at 44.1 KHz sampling rate, you only get two coordinates to ascertain that the recorded frequency is 22.x KHz, four coordinates to ascertain 11.x KHz, etc. With that little information, there is going to be sonic detail that gets lost.

Not essential, as things sound just fine at 44.1 KHz, but finer resolution can't hurt.

It can't do any good either. See nyquist theory.

The original reason for increased sample rate was to allow for gentler anti-aliasing filters. The brick wall filters necessary to tame aliasing at 44.1k were deemed to be unmusical or somesuch thing.
If the resulting numbers race makes any difference depends on whether you ask the engineering or marketing departments.

[shamann]

Fine, fine, I recant. But on a point of language, if you compare the definitions for digital sample and the word coordinate, they are the same thing. Beyond that, I accept defeat.

[duncanparsons]

sorry to rant all, please continue to share your 'golden ear' Meffy stories

While I'm at it I'll share mine, one of my projects in grad school was a nice lil audio watermarker, something that you could encode a bunch of text with (to say play back lyrics in real-time ).

Anyway to make a long story short I 'hid' most of the signal above 16kHz. Out of 20 or so guinea pings, one girl claimed to be able to hear it, and described a pinching sensation in her nose.

If you are really bored you can read about it here
http://www.ecel.ufl.edu/~jbirr/watermarking/index.html

That wasn't OneGuineaPingOnly was it?

also, cf my comment about HF and nasal sensation

DSP

[ericj23]

just to clarify - the nynquist theory requires an umlimited time frame - i.e. it is an infinite function -

so with a dac that could analyse all the samples in a song at once it would be more likely to get all the frequencies correct

if it is dealing with a few samples at a time (which it is )i'm afriad that it is more likely to make mistakes

this is 100% true go and look at the symbols used to write the nynquist thingie down -

whether this potential for mistakes intrudes into a modern dacs functioning is beyond my ability to consider

but dvd-a/sacd is a big step up from cd on my system

and of course higher frequncies harmonics can produce positive interference within a stero system

[ericj23]

http://www.digital-recordings.com/publ/pubneq.html

[Meffy]

My ear is of flesh and bone and nerve, no gold to be found anywhere near it.

[birrbits]

just to clarify - the nynquist theory requires an umlimited time frame - i.e. it is an infinite function -

I'll stop your right there.
The nyquist theory says no such thing, it assumes that functions are periodic. ie, you dont have to wait till infinity to say that a 50 Hz tone is 50Hz.

[rockstar_not]

just to clarify - the nynquist theory requires an umlimited time frame - i.e. it is an infinite function -

I'll stop your right there.
The nyquist theory says no such thing, it assumes that functions are periodic.

Thanks birrbits. I was about to go there but stopped - I only took the class, I wasn't a teaching assistant. My degree is in Mechanical Engineering, but I did ace the DSP class because it was so interesting to me. Sort of made the EE guys in the class mad.

-Scott