The Manual says, “Threshold of Clipping (%): How close to the maximum sample magnitude any sample must be to be considered clipped.” If the sample is clipped at 0 dB, that’s the highest magnitude, correct? If you only want to reconstruct the clipped portions would the best setting be 100%?
If the peaks are clipped absolutely flat and all at exactly the same level, then 100% will work.
In practice, the clipped level is often not quite flat and not quite at the same exact level, so a little bit of leeway will often produce better results. The default 95% setting is usually about right.
ClipFix does not do miracles. To restore clipped audio the effect has to guess what the missing part was like but has no way of knowing what the missing part was really like. It can be useful if just a few peaks are slightly clipped, but by the time it sounds like fuzzy distortion it is probably beyond repair.
Thanks Steve, that explains a lot.
I’ve been experimenting with some heavily-clipped files at 95% and it seems to be working better than I had expected.
You can get good results if the original show was carefully produced to be mostly level and well behaved. In that case, clipping could be only 1% or less of each wave (but look like a jungle of red clip indicators) and Clip Fix will be seen to work wonders.
Don’t try that with a live microphone recording.
Koz