Peter directed me to Julius Smith's web pages:
| see http://ccrma-www.stanford.edu/~jos/resample/ for a detailed examination
| of bandlimited interpolation.
Thank you. Unluckily, the stuff there was way over my head.
| assuming a sound with no sound components above a certain threshold (eg for
| 44.1kHz audio, nothing above 20kHz or so),
Yes, of course; a pitch higher than half the sampling rate cannot be properly
represented. I was taking it as a given that the signal would not have any
components higher than half the lower of the two sampling rates so that that
would not be a consideration, but I should have said so specifically.
| resampling to a higher sampling rate (eg 32kHz->44.1, or 44.1->48) can
| be done with almost no degradation of quality.
^^^^^^
| the original wave can be almost perfectly modeled, so
^^^^^^
| that if you then resample it to 48kHz, the sound will be almost identical.
^^^^^^
| similarly, if you have a 48kHz sound that also has no sounds above 20kHz
| [it] can be almost perfectly downsampled to 44.1kHz.
^^^^^^
| while there is a signal degradation in theory, it is extremely minimal,
^^^^^^^
Four almosts and a minimal, so there could be some. Resampling is very
faithful but not perfect.
Maybe I understand part of it now: in an actual DAC+ADC passage, analog out-
put has to come out of the DAC within the limits of its ability to generate
the detail in the voltage changes, and the ADC has to read it within the lim-
itations of its ability to sense the voltage readings; plus, as I said be-
fore, there is potential lossiness in the analog travel between them. Those
are three weak points not present in a sampling rate converter.
Nonetheless, resampling has to go from discrete samples to a representation
of a continuous waveform and then back to discrete samples, true?
| completely unlike passing through a dac->adc.
You still don't have me on the "completely" part, Peter. I do see similari-
ties. But your point seems to be that the worst part of DAC+ADC would be the
analog communication rather than the interpolation or the resampling, and a
sample rate conversion is not a dangerous thing; thank you for explaining
that.
-----------------------------------------------------------------
To stop getting this list send a message containing just the word
"unsubscribe" to [EMAIL PROTECTED]
