i found a very good wind-noise-generator example made with ‘pure data’. there a lowpass filter with a cutoff frequency at 0.1 Hz is applied. i transposed it to FMOD but get unwanted artefacts as if i set the PureData-Cutoff to 1.
So is there a workaround for FMOD getting lower cutoff or could the next FMOD release cover this?
- marc asked 12 years ago
It doesnt make sense to have a cutoff of 1hz vs 0.1hz. 1 hz is a signal that you can count once per second, and .1 is every 10 seconds. Both are totally inaudible so it wouldnt make any difference.
Which filter are you talking about anyway, there are 3 lowpass filters in fmod’s filter set.
Are you sure the cutoff frequency is measured in Hz in your example?
It might also mean 0.1 times the max frequency. For a sample rate of 44.1 khz, the max frequency is 22.05 khz, and 0.1 would mean 2205 Hz, which could make sense for wind-generation I think.
- Adion answered 12 years ago
here the wind-tutorial:
http://obiwannabe.co.uk/tutorials/html/ … under.html
i know it would be inaudible and it would be used very rarely but in that wind-tutorial the lowpass extracts a very smooth wobbling random signal out of a white noise generator afterwards its amplyfied by 80000 and taken for the midfrequency of a bandpass so the banpass is moving by random. well i could just use a randomnumbergenerator instead of filtering out the noise. doesnt matter what filter supports it – the ITLOWPASS comes nearest with lower bound of 1 Hz – so this one maybe?
no, PureData cutoff s in Hz
thanks for your replies,
- marc answered 12 years ago
The cutoff is in hz, but it depends how ‘accurate’ the filter is by the depth/order of the IIR filter.
ITLOWPASS isnt very accurate, LOWPASS is very accurate (it is an analog filter) but a bit slower, and LOWPASS_SIMPLE is fast but probably somewhere inbetween in accuracy, has no resonance so it trades that feature for a wider cutoff frequency range (it can open up all the way to 22khz – ITLOWPASS can only open up to 16khz or something i think).
I’m curious to know what measure of success you had implementing that example in FMOD? The use of a signal rate cutoff control is merely an efficiiency to save on unit generators, because we can reuse the noise code. A random number generator updated a few times a second with a k-rate interpolation would probably sound as good (as long as the line segment changes direction frequently enough). The key to realistic wind generation as described in the text is parallelism , scatter a few point sources of resonant filtered noise at three or four different average center frequencies in a scene, link them to the same control signal and you should be good to go. If you have a global wind speed parameter in world try Poisson distribution with PMF (e^-p * p^k)/k! (use a table don’t compute it) following that, and any frequencies over 2 or 3Hz are "gusts" (local fluctuations).
- Andy Farnell answered 11 years ago
Please login first to submit.