Draw multivariate Gaussian distribution samples using Python numpy.random.randn

I'm studying about Gaussian Mixture Model and came across this code which draws a number of samples from 2 bivariate Gaussian distributions. Which I don't understand is the technique that is used in the code:

import numpy as np

# Number of samples per component
n_samples = 500

# Generate random sample, two components
np.random.seed(0)
C = np.array([[0., -0.1], [1.7, .4]])
X = np.r_[np.dot(np.random.randn(n_samples, 2), C),
          .7 * np.random.randn(n_samples, 2) + np.array([-6, 3])]

(Original link: http://ift.tt/2szDwf8)

According to this Wikipedia link, we can generate multivariate Gaussian samples by Cholesky-decomposing the covariance matrix, then multiply it with a vector composed of components drawn from standard-normal distribution.

My question is the C variable in the code is not a lower triangle matrix, so how does it make sense in the multivariate Gaussian random generation?