Wave-to-Wave Recording. Noisy on Low Volume!

Curious wrote:

> It all occurs within the computer itself.

Sure, because the data is being re-quantized.

> No cables needed. The only  thing "analog" is the end product, that is the
signal the human ear
> can perceive. Before the d-a, the information is purely digital.

Not a problem, because the data is being re-quantized.


> It is poor a-d (not d-a) conversion that results in quantization
> noise.

As others have pointed out, there are many kinds of processing in the
digital domain that cause loss of SNR. Usually the processing is of
sufficient quality for noise to not be a problem. However, this is not
guaranteed.

> An analog signal too soft will have its values improperly
> rounded. This will cause quantization noise.

The same thing can happen with digital domain level changing (even slight!),
equalization, compression, a host of things.

> I think - but don't know - that the noise I'm getting is due to random
> amplification of any expected digital signal. This is a lot like a PCM
> radio receiver "trying its best" to catch and amplify a signal in an
> area:
>
> 1. With extremely poor reception
>
> AND
>
> 2. Pervaded with EMI, RFI, magnetic, and other inteferences

Name your sound card!


> The PCM receiver expects a digital signal and not an analog one. Any
> surrounding "PCM-like" radio waves at the receiver's frequency will
> cause disruption.
>
>> Also check that you don't have other
>> sources enables such as the line-in or mic inputs which will
>> contribute.

> I mute all inputs other than the "wave in"

Just guessing, but  your sound card is a SoundBlaster, right?

"Arny Krueger" <arnyk@hotpop.com> wrote in message news:<9-GdnRJuQZzBkPXdRVn-uw@comcast.com>...

> Name your sound card!

SoundMAX

Curious wrote:

   ...

> It is poor a-d (not d-a) conversion that results in quantization
> noise. An analog signal too soft will have its values improperly
> rounded. This will cause quantization noise.

Not so. Imagine that you record a single sinewave (16 bit words) with
peak amplitude $7FFF and $-7FFF. There is some round-off error, but not
enough to keep it from sounding pretty good. Now reduce the volume by 6
dB (divide all amplitudes by 2). The maxima will be $3FFF and $-3FFF.

Whoa! We started with 16 bits, and now we only have 15! To get back to
the original number, we can always just double, right? Wrong! Try it,
and you'll see that there are only even numbers in the file. Since you
know ahead of time that the LSB is zero, that's still only 15 bits worth
of information.

Call it quantization error, call it truncation error, call it whatever
you like. Once you discard information by lowering the volume, you can't 
get the information back.

   ...

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

curious11112001@yahoo.com (Curious) wrote in message news:<34a4f456.0403290055.2ca6aab3@posting.google.com>...
> chris@nospam.com wrote in message news:<iace60hkrbgkdfo2fvcmaqvdbuav91b7c8@4ax.com>...
> > Please describe how you are doing this re-recording? 
> 
> Play wave file in Realone player and record using Wavelab at approx.
> the same time. First click record in Wavelab and then click play on
> Realone.

So, in other words, you don't know if it is routed back through the
cound card or not.

> > If this is
> > windows and you're playing the file and using a recorder, then you are
> > actually doing a/d -> d/a.  
> 
> It all occurs within the computer itself. No cables needed. The only
> thing "analog" is the end product, that is the signal the human ear
> can perceive. Before the d-a, the information is purely digital.
> 
> It is poor a-d (not d-a) conversion that results in quantization
> noise. 

NO IT IS NOT. It is ANY process, digital or analog, that results
in requantization. You've been given several examples and explanations
where even a pure digital process MUST result in quantization noise,
and the means necessary to alleviate the problem.

> An analog signal too soft will have its values improperly
> rounded. This will cause quantization noise.

So will ANY process that results in requantization. Do you
understand that?

> I think - but don't know - that the noise I'm getting is due to random
> amplification of any expected digital signal. 

Nope.

> > Also check that you don't have other
> > sources enables such as the line-in or mic inputs which will
> > contribute.
> 
> I mute all inputs other than the "wave in"

And, thus far, you have not been able to prove that the "wave in"
recording is NOT routed by the relevant device drivers to the D/A 
converter of your sound card and routed right back in through the
A/D conversion. No external cables are needed, the switching can be
done right on the soundcard itself.

Curious wrote:
> "Arny Krueger" <arnyk@hotpop.com> wrote in message
> news:<9-GdnRJuQZzBkPXdRVn-uw@comcast.com>...

>> Name your sound card!

> SoundMAX

You understand that this  is a low-end on-motherboard consumer sound card,
right?

Arny Krueger wrote:
> Curious wrote:
>> "Arny Krueger" <arnyk@hotpop.com> wrote in message
>> news:<9-GdnRJuQZzBkPXdRVn-uw@comcast.com>...
>
>>> Name your sound card!
>
>> SoundMAX
>
> You understand that this  is a low-end on-motherboard consumer sound
> card, right?

No doubt this is the ADI 1981 Soundmax, currently used for on-board sound on
a large number of motherboards.

The basic doc for this chip can be found at
http://www.analog.com/UploadedFiles/Data_Sheets/35206112273846AD1981B_b.pdf
.

Someone can correct me if I've overlooked something, but according to the
"Functional Block Diagram" , the chip has only a digital SP/DIF output, and
no discernable digital inputs. All other inputs and outputs are analog.

According to page 3  the ADC dynamic range is 85 dB A weighted with
nonlinear distortion down about 84 dB down, while DAC dynamic range is 85 dB
with line output nonlinear distortion 85 dB down. This is really pretty good
for a low-end on-board chip, but it is insufficient to guarantee anything
like bit-perfect performance.

All operations that involve copying computer data through this chip can be
expected to be measurably degraded, with the loss of  roughly the lowest two
bits worth of resolution.

>>> SoundMAX
>>
>> You understand that this  is a low-end on-motherboard consumer sound
>> card, right?
>
>No doubt this is the ADI 1981 Soundmax, currently used for on-board sound on
>a large number of motherboards.
>
>The basic doc for this chip can be found at
>http://www.analog.com/UploadedFiles/Data_Sheets/35206112273846AD1981B_b.pdf
>.
>
>Someone can correct me if I've overlooked something, but according to the
>"Functional Block Diagram" , the chip has only a digital SP/DIF output, and
>no discernable digital inputs. All other inputs and outputs are analog.

Sure looks that way to me.

If I'm reading this correctly, the "record what I am playing back"
signal path goes through the 20-bit sigma-delta DAC (and, possibly,
through the signal-rate-conversion logic) and converted to analog. The
analog signal flows through a gain-adjust and muting block, through an
analog summing block, and is then fed back up to the record selector,
through another gain-control block, and into the 16-bit sigma-delta
ADC.

I do not see a purely digital "record what I am actually hearing" data
flow path.

-- 
Dave Platt <dplatt@radagast.org>                                   AE6EO
Hosting the Jade Warrior home page: http://www.radagast.org/jade-warrior
  I do _not_ wish to receive unsolicited commercial email, and I will
     boycott any company which has the gall to send me such ads!

Curious wrote:
> Why is it whenever I record from one wave file - at an extremely low
> volume - another wave file, I get noise?
>
> The process is completely digital. There is no analog, no D-A, and no
> A-D. Why the #%!$@%!@# is their noise?
>
> Since this is not analog-to-digital, this is obviously *not*
> quantization noise. Quantization noise only occur during A-D
> conversion when the input volume and/or bit-resolution is inadequate.

"Record from one wav file to another "  - what ?  What's wrong with
copy/paste ?

geoff

PS Dunno how you could be getting your noise, but considering you are
attempting something a bit whacky in the first place (maybe), anything's
possible.