by Mr_-G- » 18 Oct 2015, 13:33
One should not be jumping the start loop from here to there as shown in that video, hoping to find 'by chance' a good loop point. It just won't happen.
Here is a list of things I do, I hope is useful to others:
0. I normalise all the samples (in Audacity) before loading into the NSE. If there is a noticeable background presence, I sometimes also remove noise with the noise reduction command in Audacity.
1. In general, looking a the low resolution of the wave, you can sometimes see large scale cyclic variations of amplitude and those give a good indication of what the loop region could be.
Most of the beating in the loops are because these cycles (phases and the amplitudes) do not match and therefore you end up with a different spectral contents in 2 parts we are trying to mix.
2. If the sound has a sharp transient, like a piano sound, do not start the loop too close to the start of the sound as it will be practically impossible loop seamlessly.
3. Start with no crossfade (XFade) at all, and try to match the red and green waveforms in the right side pane, using first the Length Coarse and then Fine Phase adjust controls, but do not alter the Loop start.
4. Only then use the XFade to see how that mixes up. Long XFades are no solution to badly matched loop points.
I found many times that the XFade does change the colour of the sample (phase cancellation and all that), so use it cautiously and aim for short XFades first.
I often found that the 3 XFade Curves did not make much of a difference and tend to use the Linear (default) more often.
5. If fiddling with this does not yield any good results only then change the Loop start and restart the whole process again.
6. When re-sampling a module that also uses looped samples, chances are that you might be able to find some good looping points, so recording sample longer that you might end up using is perhaps a good idea.
7. When sampling wind instruments that do not seem to have a lot of variation in time (like a flute, oboe, etc.) you might want to use a long loop first anyway, because there are tiny changes in pitch and tone that make a difference to the natural sounding of the sample. If you loop too short, (a few cycles) it will sound synthetic. One way to check for this is with Audacity, showing the spectrogram of the recorded track rather than the waveform.
Enjoy.
Last edited by
Mr_-G- on 18 Oct 2015, 18:24, edited 1 time in total.
- The author Mr_-G- was thanked by 5 members, including:
- Leo Castro • Berretje • oyarsa • pablomastodon • tomzi