QUOTE (precisionist @ Jan 25 2005, 07:11 AM)
1. How is it possible that a sample rate conversion process could cause clipping ?
2. The 'true' analogue signal may go beyond 0dB, but none of the 48kHz samples does and the 44,1kHz samples also won't, since they're the same bitdepth (16bit, I suppose).
3. 'True' analogue signal above 0dB could in principle sound like clipping (but it's surely not audible, because too faint), but isn't actually "clipping", since no samples are clipped. Only samples are relevant for representing the audio data.
1.
Here's an example: A 12 kHz sine at a sampling rate of 48 kHz can look like this:
+1 +1 -1 -1 +1 +1 -1 -1 +1 +1 -1 -1 +1 ...
this sine has an amplitude of actually 1.4142... If you sample this sine at another rate like 44,1 kHz you'll get clippings. (+3 dB)
2. see my previous paragraph for a counter example
3. Usually a D/A conversion involves upsampling. You should always have some dynamic headroom for this. I can't tell you how it's usually done excactly or if the effect you describe is actually an issue here.
SebastianG
edit: spelling