|
 Kyma Forum
 Tips & Techniques
 extreme DC drift
|
| next newest topic | next oldest topic |
| Author | Topic: extreme DC drift | |
|
phillipm Member |
posted 25 June 2015 01:30
 
IP: Logged | |
|
pete Member |
posted 25 June 2015 07:19
This is the beauty of Kyma in that it gives you what you ask for and not what it thinks you want. If you have a sample cloud and it just so happens that the part if the waveform you are reading is at the lower part of the cycle you will get a DC off set, especially if you have a large quantity of the same bit being played at the same time. It could have been made with DC filters in the players so that they don't accumulate DC, but then it would limit what you could use the module for. You may actually wanted the accumulated DC and treat it as a control signal for example. Instead you need to put in your own DC (hi-pass) filters to stop it or hi pass filter the the original sample that is used for the sample cloud so that it is less likely that the window will fall on a point with the DC offset. If needs be you could make the original signals quieter so that you can filter the result (mix of grains) before boosting the gain at the output. I think it will take quite a lot of attenuation before you hear any degradation in the signal. IP: Logged | |
|
SSC Administrator |
posted 25 June 2015 08:33
Pete makes a good point about the wavetables. Do all of your samples and wavetables end with a zero value? If archive and upload your Timeline, we can look for the source of the DC offset. IP: Logged | |
|
phillipm Member |
posted 25 June 2015 09:28
Thanks Pete, I should have thought of this last night in the midst of the hour or so that I was fumbling around with levels and random seeds on a 13-track TimeLine trying to keep Kyma from clipping. :-( Just double-checked the timeline and this one is actually all GrainClouds. No samples were used other than three of the basic waveforms located in the "Waves" folder sine, saw1024, cycloid2. Those last two clearly do *not* end with a zero, so is that the *source* of the problem? So I'm assuming the optimal placement of the high pass filter is post-transposition (after the GrainClouds has generated all those grains. A quick test using a HPF with a 20 Hz cut-off level things out and increased headroom by c. 35% without a noticeable change in audio content (at least not on a quick superficial listen).
[This message has been edited by phillipm (edited 25 June 2015).] IP: Logged | |
|
SSC Administrator |
posted 25 June 2015 15:14
Yes, a 10 Hz cutoff should be sufficient for the high pass filter. Thanks for sending the Timeline. The Cycloid2 waveform itself has a negative DC offset. It isn't balanced around zero (plus it has a huge discontinuity between the end of one cycle and start of the next which results in aliasing because there are multiple cycles within each Gaussian envelope). As you increase the levels, you are multiplying all the DC constants by a larger number with the result that the entire waveform gets more and more shifted below zero. IP: Logged | |
|
SSC Administrator |
posted 25 June 2015 15:23
• Select all Then save the new cycloid as CycloidNoDC, that should remove or greatly ameliorate the problem of the DC offset. An example of the new waveform is attached to this post. IP: Logged |
All times are CT (US) | next newest topic | next oldest topic |
  |
|
Contact Us | Symbolic Sound Home
This forum is provided solely for the support and edification of the customers of Symbolic Sound Corporation.
Ultimate Bulletin Board 5.45c