Crossfading

Note that products can be fed backward also note that crossfades are found in ROMs which are input to just one side of the multiplier the inputs including faders and audio inputs are fed to the other side after been converted via logarithmic ROMs. 

Crossfade looping. In many cases, a perfect loop is difficult or impossible to find. It can be helpful to perform what is called a crossfade loop. Here the loop data are modified to produce an acceptable loop with a signal processing rather than a signal analysis operation. 

The basic principle of a crossfade loop is to average data from the beginning and the end of a loop. If the waveform data at the each end of a loop are replaced with the average of the data from the both the beginning and the end of the loop, the splice point will be inaudible. This is a simple operation denote the loop start as x[s] and the loop end as x[e], where s denotes the loop start offset and e denotes the loop end offset, then we have the modified waveform data... 

While the sampler is playing data back from within the loop itself, there will be no click at the loop splice point. Of course, as the sampler read pointer crosses the transition between unmodified waveform data into the modified (averaged) data, there potentially will be a discontinuity. So the second element of crossfade looping is the... 

The process of a backwards forwards crossfade loop does produce artifacts if the signal has significant anti-symmetric (odd) components around the loop point, so it is best performed on a loop point that has already been screened for a close fit to being symmetric. 

Relation of Crossfade Looping to Time and Pitch Scaling. The operation of crossfade looping is essentially the same operation that is performed in pitch or time scaling schemes as described in the chapter by Laroche except that in sampling the operation is typically performed once, off line, and with pitch or time scaling, the operation is performed in real time continuously, in order to scale pitch independently from time. 

A n can be any number of window shapes, but the two most commonly used are a triangular window, and a cos2 window. In both cases, the window should sum to one when overlapped with itself by W-72 samples, and should be unity at the center when the waveform segments that are overlapped correlate highly. If the overlapped waveform segments have a very low correlation, then it is preferable to use an equal power crossfade, where the sum of the windows is 1.414. This strategy tends to keep the RMS energy close to constant across the cross fade. 

Backwards Forwards Crossfade Looping. A backwards forwards loop at any arbitrary point can also be created (or improved) by modifying the loop data. Here the crossfade is performed not between the data at the beginning and end of the loop, but from before and after the loop points. 

The objective of a backwards forward crossfade is to modify the data at the loop so that it is a purely even function around the loop point. Thus we simply replace the data at the loop point with the even part of the function, and fade back into the original data going away from the loop point itself. 

Crossfading is an approach which combines elements of multiple wavecycling and sampling. This works by gradually changing the sample outputs from one wavetable by the... 

Figure 5.19 An overlap crossfade window is normally used to avoid discontinuities between the segments in the brassage process...

---