Issues 'playing' aerodrums.

InTheWorks
Posts: 80
Joined: Sat May 28, 2022 11:59 pm

Re: Issues 'playing' aerodrums.

Post by InTheWorks » Thu Jun 23, 2022 7:17 pm

Richard wrote:
Thu Jun 23, 2022 12:29 pm
With Aerodrums though, there really isn't any need to use big swings like this and we recommend that you're more economical with your motions.
This is a good point and I suppose will come with time and practice.
Richard wrote:
Thu Jun 23, 2022 12:29 pm
It would be interesting to try the approach with "continuous openness" where the hihat sample constantly fades between different samples depending on the current position of the hihat pedal.
I started with these not very nice hihat samples. They are all the same velocity, but different openness. And wrote a matlab script to cross fade them. They all have the same cross fade period and in this picture it's 50ms. So the fade in is 50ms, there's 50ms of audio, then 50ms of fade out. The fade in is not applied to the first hihat sample and the fade out is not applied to the last hihat sample. Just to be clear.
Screenshot at 2022-06-23 11-01-39.png
Screenshot at 2022-06-23 11-01-39.png (119.69 KiB) Viewed 4800 times
Hopefully it's obvious that the timescales are not the same for the left and right half of that image.

The original
full-open.mp3
(111.67 KiB) Downloaded 283 times
and the combined cross fades
crossfade-50ms.mp3
(59.81 KiB) Downloaded 281 times
I tried shorter periods, but it wasn't as noticeable.
Richard wrote:
Thu Jun 23, 2022 12:29 pm
We do have old implementations of these ideas so it may be possible to get you a test build. I will need to look into it.
If you can, I'd love to test it.

Richard
Posts: 1012
Joined: Wed Feb 26, 2014 1:45 am

Re: Issues 'playing' aerodrums.

Post by Richard » Fri Jun 24, 2022 10:22 am

Nice, thanks for doing that test. The version with cross fades definitely doesn't sound unnatural in any way, to me.

What I expect we will try is to have samples like yours and then, given the current hihat openness level, find the two samples that are closest to that openness level and take a linear combination of them. Then repeat this selection of new samples at a high frequency (e.g. every 10ms).

We will let you know the results when we do it.

Post Reply