|
 Kyma Forum
 Tips & Techniques
 Nyquist and DC filter on Paca
|
| next newest topic | next oldest topic |
| Author | Topic: Nyquist and DC filter on Paca |
|
cristian_vogel Member |
posted 17 February 2010 15:36
 
What would be the best approach to make an ideal filter that can cutoff DC below 20 hz and stop frequencies above SignalProcessor halfSampleRate on the Paca(rana) ? IP: Logged |
|
SSC Administrator |
posted 17 February 2010 17:18
A high pass filter can be used to block frequencies below 20 hz. Or you can subtract the Sound from itself delayed (the changing delay will change the cutoff frequency). As you can imagine, subtracting a Sound from itself delayed will wipe out DC, because the DC value is the same no matter what the delay. It is impossible to get any frequencies above the Nyquist limit, so there are no frequencies above the half sample rate. For example, if your sample rate were set to 44,100 hz and you tried to synthesize the frequency 33,100, the frequency you would actually hear would be (44,100 - 33,100 = 11,000 hz). IP: Logged |
|
cristian_vogel Member |
posted 18 February 2010 01:02
the cutoff filter method should be a first order filter? I want to cutoff at Nyquist exactly because I would like to block those foldown frequencies. They can arise from samples that are being played back at a much higher frequency than they were recorded at. For this lowpass filter, what would be the closest to an ideal filter model? IP: Logged |
|
pete Member |
posted 18 February 2010 09:48
Hi Cristian You would have to filter out the higher frequencies first, before you up the pitch, else it's too late. The frequencies will have already folded and been mixed in with the sound so no filter could separate them out at this stage. What you would need to do is calculate what the final nyquist point will be on the sample before you've changed the pitch. That is if the final signal was two times the original pitch you would filter out frequencies above half the nyquest (1/4 sample rate) on the original signal before you repitch it to the new higher signal. This is not so easy to do because the new signal will be playing through the sample twice as fast as the original signal so you would need to make a special filter that could process the samples at double the sample rate (or more if the output pitch was higher). Alternatively you could pre process the sample and make pre filtered copies, one for each of the possible desired final frequencies. The Surbiton oscillator calculates what the final pitch is going to be and filters the wave form before re-pitching happens. It also suffers from the problem that the original pitch takes longer to process than the final output but because it is looped it soon catches up (within a few cycles). I hope this makes sense Pete IP: Logged |
|
SSC Administrator |
posted 18 February 2010 11:57
Another alternative would be to do a live spectral analysis/resynthesis of your sample. You can leave the sample at its original frequency and instead change the frequency of the resynthesizing oscillators. The OscillatorBank automatically filters out the oscillators whose frequencies would go above half the sample rate. I would recommend using the non harmonic version of the live analysis for this application (re-pitching without any morphing). 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