How do I use the "mle" function for a custom distribution?

I have a binomial measurement vector t=(0:tao:T) (tao and T are known) of length 10^5 comprised of ones and zeroes. The probability for getting a 1 in an element t is

p = sin((omega/delta).*(sin(delta.*(t+tao)+phi)-sin(delta.*(t)+phi))+pi/4).^2

where omega, phi and pi are known constants, t and t+tao are elements in the vector and delta is the unknown parameter.

I would like to use the mle function of the "statistics and machine learning" toolbox to discover the maximum likelihood of delta, but the documentation is not exactly clear on what I am suppose to do in it. Specifically, what exactly is nloglf and how do I define it?

Thanks in advance to all the helpers.