All instruments and drums again are mathematical expressions. This started off as another Xpressive test where I was creating new timbres which morphed into something inspired by and resembling the bassline from It's More to Compute by Kraftwerk. The saw tooth was also inspired by Numbers by Kraftwerk.
This originally contained vocals which were 4 bit piecewise constant segments "sampled" at 1000 Hz. I wished to show that I had worked out how to consistently convert samples to mathematical expressions without crashing (MATLAB performed the calculations). I worked out that as we are working in continuous time and not discrete time, a ternary expression such as (condition ? value_if_true : value_if_false) must be used. The last term must have a value for value_if_false which can be 0 or the DC offset correction, the other terms should have no value. There can be a total of 1060 segments before crashing, which I further worked out can be nested to add more segments. I had actually managed to "sample" at 8 bit 16,000 Hz. However, the file size was getting large as you can see with the size of this file with just the 4 bit 1000 Hz "samples". I have left the vocals in but greyed out. I wasn't sure whether to leave them in out. I didn't want to get a one star because they sound terrible at 4 bit 1000 Hz.
My experiments, including those detailed in my first Xpressive "tutorial" post, indicate that while Fourier series and amplitude segments can technically represent vocal samples in Xpressive, they are not feasible due to their impact on file size and CPU performance. Also, with sampled synthesised instruments such as stabs, it has become apparent that I have been getting better results preforming an FFT and expressing all (or top n) frequencies instead of sticking to strict Fourier series of (n) integer harmonics.
This is not a proper cover, as it is not a true or close recreation of the original. Again, like the others feel free to use any of the instruments copyright free.