I pass on this interesting item by Peter Stern from the Editor's choice section of Science Magazine:
Even the best musicians make slight errors when playing a rhythm. We find this frailty to be appealing, as evidenced by the fact that computer-generated perfect rhythms are often perceived as sterile or artificial. Having known this phenomenon for a long time, software engineers have added slight rhythmic fluctuations to make computer-generated music sound more human. These fluctuations are usually produced by a random number generator. Hennig et al. have now analyzed the statistical properties of music produced by professional musicians. They found that there are long-range fluctuations when humans produce all sorts of rhythms. A small rhythmic fluctuation at some point in time not only influenced fluctuations shortly thereafter, but even after tens of seconds. When given the choice, listeners clearly preferred music produced according to these criteria over the random number-generated fluctuations. The authors conclude that these results may not only have practical implications such as improved techniques for audio editing and humanizing music, but they may also provide new insights into the neurophysiology of time perception and timing of actions.
Here is the
Hennig et al. abstract:
Although human musical performances represent one of the most valuable achievements of mankind, the best musicians perform imperfectly. Musical rhythms are not entirely accurate and thus inevitably deviate from the ideal beat pattern. Nevertheless, computer generated perfect beat patterns are frequently devalued by listeners due to a perceived lack of human touch. Professional audio editing software therefore offers a humanizing feature which artificially generates rhythmic fluctuations. However, the built-in humanizing units are essentially random number generators producing only simple uncorrelated fluctuations. Here, for the first time, we establish long-range fluctuations as an inevitable natural companion of both simple and complex human rhythmic performances. Moreover, we demonstrate that listeners strongly prefer long-range correlated fluctuations in musical rhythms. Thus, the favorable fluctuation type for humanizing interbeat intervals coincides with the one generically inherent in human musical performances.
A friend of mine comments:
ReplyDelete"I personally can't tell if a rhythm track has been quantized or humanized when listened to it in context (as a whole song). A quantized rhythm track using bad samples sounds like a machine gun and humanizing the track doesn't really fix the issue. Better samples with a large round-robin bank usually the medicine for that problem.
For Guitar Hero I handled the load of "tempo mapping" for the songs we put on the disc. Since most of those songs were recorded before the invention of DAWs and many before the incorporation of click tracks I can safely say that 90% of the songs I worked on were "humanized" naturally by the artists. When listening to the whole song, I couldn't tell if a Black Sabbath song from the 70s or a modern Maroon 5 song were humanized or not since the whole band is "in the pocket", but when I would isolate the rhythm tracks and in most cases the drums I could tell a quantized track was quantized but magically the old "natural" track (Black Sabbath) just sounded…natural. Again if the drums were sample based and they didn't use a broad library then each repeating drum stood out, ick.
I have used the humanize feature many times in the past and found that less is more. Too much stands out and too little is not noticeable. Also the tail of the sound becomes an issue when humanizing. Snare quantized to the grid work pretty well since they don't have much of a tail (dry of course) where as a boomy kick drum snapped to grid my cut off the tail of the previous kick drum and cause the ear to say, "I heard something weird". That said, I've only humanized if I felt the piece isn't in the pocket and then, I do it very subtly to make sure no tails get cut off and the pocket isn't lost. Orchestral sample libraries is where I apply this philosophy the most."