rlw@ttardis.UUCP (Ron Wilson) (01/04/90)
In article <523@lexicon.com>, fc@lexicon.com (Frank Cunningham) writes: >In article <1989Dec31.055208.2339@smsc.sony.com> dce@smsc.sony.com (David Elliott) writes: > >> So, I assume that most multi-effects units consist of a general purpose >> processor, some special-purpose processors, and a bunch of general use >> RAM. >> >> Is this correct, or do they use multiple general-purpose processors? > >A typical multi-effect unit may consist of a general purpose processor >and its ROM/RAM space to handle the user interface and internal >house-keeping and usually 1 special purpose processor (usually a >proprietary DSP) for doing the audio. It will typically have a 64k >audio RAM which means over a second if used as a pure delay, depending >on the sample rate. > >Pitch shifting involves a complex real-time interaction between the GP >and the DSP. > >Reverb and Delay are RAM intensive, but only require GP/DSP >interaction when a parameter is changed. > >EQ and feedback don't require much RAM. > >So the complexity of your multi-effect is a resource allocation >trade-off. > >> Also, how do they get the sound to come through without much perceptible >> delay? > >All these effects are time-domain based. There are no frequency-domain >processors that you'd care to pay for. There is a small perceptible >delay through a typical pitch changer. How does one do EQ in a time domain? The only digital method I've ever heard of for doing EQ required transformation to and from the frequency domain (but then I concentrated on computer EE - so my general EE knowledge is a bit weak)