BLITs and minBLEPs (Audio synthesis stuff)

Status
Not open for further replies.
M

mattomatto

Well-known member
Hi all,

I'm using the Teensy to develop the best quality synthesiser I can, learning all about digital synthesis on the way. The wavetable method is all well and good but all the cool kids seem to be using minBLEPS with very clean results and I want in!

Most of the info I can find relate to software implementations and I was hoping someone here could offer some advice as to whether this could be the way to go with the Teensy? Is it CPU efficient enough?

Many thanks.

Matt
 
Well, I'm not really the best person to explain.. but as I understand it, the basic premise is to solve the problem of oscillator aliasing by replacing the problematic parts of a 'naive' waveform with pre-calculated minimum-phase bandlimited steps, or minBLEPs. Mixing these minBLEPs in is costly meaning the higher the frequency, the more CPU is required.

The maths and code is a little out of my league right now, but if it's possible, I'd like to start learning :)
 
Its also worth noting that this is not aimed at general waveforms. Its specifically about producing inherently bandlimited rectangular pulse trains, such as square waves and related forms.
 
PaulStoffregen said:
Are there any fixed point implementations published?
Click to expand...

Not that I've been able to find. As it's currently pretty 'industry standard' to use this method in synthesis and a lot of hard work goes into doing it, there's very little in the way of source code and help.

The alias-free digital synthsesis paper here is a very good guide to bland limited impulse trains, which are the unpinning theory behind minBLEPS.
https://ccrma.stanford.edu/~stilti/papers/

A few guys on KVR forum say saw, triangle and sines are all possible with a bit of work.
 
Status
Not open for further replies.
Back
Top