Gianni Pavan wrote:
>I create a 16 bit stereo .wav file and I fill it
>with integer samples with values
>-32000,+32000,-32000,+32000 ... with a program
>written in Visual Basic. Such a sequence of
>samples corresponds to a perfect sinusoid at the
>Nyquist frequency (half of the sampling rate)
>that can't be recorded from an analog source.
>If you feed a non-resampling digital input the
>values are just kept and stored in a new file.
>But if the digital input device resamples the
>data stream all those value are processed with a
>resampling algorithm that produces a new series
>of samples that don't match with the original
>ones. A resampler keeps the original stream,
>upsamples to a much higher frequency, performs a
>digital filtering to cut frequencies higher than
>the new sampling rate (that could be the same of
>the input or a different one) and also attenuates
>frequencies close to the Nyquist frequency, then
>downsamples to the final rate. If your original
>stream is 48k and the output of the resampler is
>exactly 48k again, frequencies higher than
>22-23kHz will be attenuated and the "Nyquist
>signal" will be greatly attenuated.
>In some cases the input stream has not exactly
>the rate of the sound board clock and in this
>case the "Nyquist signal" will appear at a
>frequency that is not the Nyquist frequency of the new file.
>
>A "no resampling" digital input is able to lock
>on the incoming rate and to transfer samples
>without any further processing. If you open with
>Audition the file created by the digital input
>device you see very clearly the difference among
>resampled and not-resampled "Nyquist signal"
>
>I hope I was clear enough,
>Gianni
That's a cool tool. I don't have any way to make it. Could you put a
test file in a place where people could download it?
-Dan Dugan
________________________________________________________________________
________________________________________________________________________
|