The beats of a song are not always located exactly at peaks visible in the waveform, an interesting little fact that even experienced video editors are often not aware of. In this tutorial you'll learn the details and how BeatEdit for Premiere Pro can help.
From time to time we get feedback for our automatic beat detection tool BeatEdit for Premiere Pro where users report that the beat detection is not accurate. When we investigate those issues, it turns out that the users expect the beat markers to be located exactly at peaks of the audio waveform. However this is not the case - for a good reason!
The beat is a 100% regular pattern or in other words, the time from one beat to the next one stays always the same, unless the tempo of the beat changes. However, for the peaks in the waveform there are many reasons why this is not the case. First of all, any music that is based on the live performance of a musician will never be 100% accurate. It is simply impossible to play an instrument and hit each beat without being off a few milliseconds. But much of the music we hear today is based on quantized midi data where the timing in theory should be 100% accurate. Does this mean that for those songs the beats perfectly lineup with peaks? No! Midi instruments are often based on sample libraries consisting of hundreds or thousands of recorded audio samples. And inside of those samples, again the peaks are often not aligned accurately.
Is this a problem? Not at all - having peaks slightly behind or before the actual beat does not sound odd. It is barely noticeable and maybe even adds some character to the song. The important thing to learn here is that you should not be irritated when the beat is located slightly before or after a peak. If you would shift the beat markers produced by BeatEdit to the nearby peaks, your result is not getting better, but worse: Because then the beats are not distributed evenly anymore. And this is what makes up the beat: That it is an accurate, regular repeating pattern.