Originally Posted by arnyk
The above would appear be just another example of using obviously erroneous analyses to justify obviously erroneous conclusions. Of course data can be sliced and diced so that its analysis produces an irrelevant result. We've got a number of papers that give representative numbers for the crest factor of music, and none of the above correlates with them. Let's hear it for peer review!
Thank you for your perspective but I am puzzled by what you are saying. I showed a real piece of music and its real peak. I explained how one set of statistics was correct and another that was not. The correct value showed a crest factor less than 2 db -- a far cry from 10 db number you said music load imposes on the amplifier.
I took your lead and used Audition for this analysis. As such, the numbers are not generated by me at all. I showed you the actual output of the program.
I love to see what papers you have on this topic as the one I post last night does not apply because it is relying on averages, not peaks:
Your amplifier doesn't peer into your music shelf to decide what to do next. It plays whatever instant voltage is in the track being played. If it runs out of juice in one of these peaks, it matters not that if it had averaged that with 40 CDs you are not playing, the level would be lower. Please read the paper including its summary to make sure you know how to use the results out of it. This is the abstract:
Our discussion and goal is to find out the instantaneous music crest factor
because that is what causes clipping at that moment. The paper and their study is using *averages* which is fine for their application of determining speaker power ratings and such but is of no value to our current discussion.
I would be delighted to see authoritative data shows no music peak has crest factors less than 3 db. So by all means, please share the papers.