Clipping... let's dumb it down.
Reply #9 – 2005-12-07 19:35:26
2. NOT during the ENcoding-phase: ... 3. During the DEcoding-phase(playback): ... . This says the encoding process can increase the signal level above that of the wav source, at least at some peaks, but these will not clip. If true, encoding must create numbers greater than the maximum 16 bit value, and these numbers can be stored in the mp3, no? Is there a reference for this information? Correct. I am quite certain references about this are abundant, but i'm currently too lazy to look it up. Among other things, you can check it yourself - all you need is a replaygain-capable app which can display the peak values. I own tracks which virtually screw LAME to such an extend, that the peak value will be 160%! Such a value couldn't exist, unless: A. it was introduced during encoding... and B: it could be stored in the ENcoded will without clipping.If decoding causes clipping because of those higher values created in the encoding process, how could any program correct that post decoding? By scaling it down during decoding BEFORE it is put into the target output format. For specific technical details, someone else will need to help out.If item #2 in your post is correct, then how is it that I can rip a track from a CD, then have it analyzed by mp3gain and find that it's clipped? No playback involved. The "clipped" marker in mp3gain does *not* relate to what is ENcoded in the mp3, but instead what would happen if it would be DEcoded at the given volume. - Lyx