Harriman 220a

HW4


All of the data sets used were found in the Time Series Data Library.


Data sets :

daily Stock data IBM and S&P 500 from 1980-1992

these are the main voices in the front left and right

schizophrenic chemical levels during drug therapy

this is used to control FM modulation amount

Capexp and capexg

gross and final capital expenditure data from Australia from 1989-1990

These are changing more slowing in the rear channels


empper0

This is the number of employed persons in Austrainly from 1978-1991

This data is used to control the speed at which other changes occur. Other time changes are scaled based on this value.


I used some different sections to isolate a voice or two but I'm not sure this mapping makes the most sense for what I've done. I seemed to not get an update on the data file since that part of the while loop wasn't encountered. All the same, it accomplishes the requirements and has some interesting changes.


hw4.wav

Harriman_hw4_b.ck

Harriman_hw4.ck

data_files.zip


Why are the values for gain squared?

Since the same parameter is used for gain and frequency, as frequency decreases the gain will decrease as an exponential. I think this is because of the loudness curve we perceive as loudness in our hearing, lower frequencies seem louder to us, but the loudness curve goes back up at higher frequencies and these programs won't mimic this contour.


I suppose it more closely follows how we perceive loudness.


Why use MIDI keynums for the freq values?


This will map linear values to a logarithmic frequency scale which is more natural to how we perceive pitch.


What happens if they're mapped linearly, instead?

Depending how it's scaled a similar range could be achieved but the scaling of notes from a arbitrary dataset along that range will be different. In general a linear mapping has the effect of appearing (to our ears) that it is changing much more slowly at higher frequencies.


And (3) what happens if the 100 ms update rate is increased or decreased significantly?

At extreme cases where the rate is changed at audio sampling frequencies tones are heard at those frequencies. At very slow rates the pitches are simply sustained for longer periods.